Merge branch 'master' of ssh://nickel.broadinstitute.org/humgen/gsa-scr1/gsa-engineering/git/unstable

public/java/src/org/broadinstitute/sting/gatk/GenomeAnalysisEngine.java

@@ -215,8 +215,8 @@ public class GenomeAnalysisEngine {
             resetRandomGenerator(System.currentTimeMillis());
 
         // if the use specified an input BQSR recalibration table then enable on the fly recalibration
-        if (this.getArguments().RECAL_FILE != null)
+        if (this.getArguments().BQSR_RECAL_FILE != null)
-            setBaseRecalibration(this.getArguments().RECAL_FILE);
+            setBaseRecalibration(this.getArguments().BQSR_RECAL_FILE);
 
         // Determine how the threads should be divided between CPU vs. IO.
         determineThreadAllocation();

public/java/src/org/broadinstitute/sting/gatk/arguments/GATKArgumentCollection.java

@@ -192,7 +192,7 @@ public class GATKArgumentCollection {
      * and the raw empirical quality score calculated by phred-scaling the mismatch rate.
      */
     @Input(fullName="BQSR", shortName="BQSR", required=false, doc="Filename for the input covariates table recalibration .csv file which enables on the fly base quality score recalibration")
-    public File RECAL_FILE = null; // BUGBUG: need a better argument name once we decide how BQSRs v1 and v2 will live in the code base simultaneously
+    public File BQSR_RECAL_FILE = null; // BUGBUG: need a better argument name once we decide how BQSRs v1 and v2 will live in the code base simultaneously
 
     @Argument(fullName="defaultBaseQualities", shortName = "DBQ", doc = "If reads are missing some or all base quality scores, this value will be used for all base quality scores", required=false)
     public byte defaultBaseQualities = -1;

public/java/src/org/broadinstitute/sting/gatk/walkers/recalibration/ContextCovariate.java

@@ -26,6 +26,7 @@
 package org.broadinstitute.sting.gatk.walkers.recalibration;
 
 import net.sf.samtools.SAMRecord;
+import org.broadinstitute.sting.utils.recalibration.BaseRecalibration;
 
 import java.util.Arrays;
 
@@ -35,19 +36,21 @@ import java.util.Arrays;
  * Date: 9/26/11
  */
 
-public class ContextCovariate implements Covariate {
+public class ContextCovariate implements ExperimentalCovariate {
 
     final int CONTEXT_SIZE = 8;
     String allN = "";
 
     // Initialize any member variables using the command-line arguments passed to the walkers
+    @Override
     public void initialize( final RecalibrationArgumentCollection RAC ) {
         for( int iii = 0; iii < CONTEXT_SIZE; iii++ ) {
             allN += "N";
         }
     }
 
-    public void getValues(SAMRecord read, Comparable[] comparable) {
+    @Override
+    public void getValues( final SAMRecord read, final Comparable[] comparable, final BaseRecalibration.BaseRecalibrationType modelType ) {
         byte[] bases = read.getReadBases();
         for(int i = 0; i < read.getReadLength(); i++) {
             comparable[i] = ( i-CONTEXT_SIZE < 0 ? allN : new String(Arrays.copyOfRange(bases,i-CONTEXT_SIZE,i)) );
@@ -55,8 +58,8 @@ public class ContextCovariate implements Covariate {
     }
 
     // Used to get the covariate's value from input csv file in TableRecalibrationWalker
+    @Override
     public final Comparable getValue( final String str ) {
         return str;
     }
-
 }

public/java/src/org/broadinstitute/sting/gatk/walkers/recalibration/CountCovariatesWalker.java

@@ -41,6 +41,7 @@ import org.broadinstitute.sting.utils.collections.NestedHashMap;
 import org.broadinstitute.sting.utils.exceptions.DynamicClassResolutionException;
 import org.broadinstitute.sting.utils.exceptions.UserException;
 import org.broadinstitute.sting.utils.pileup.PileupElement;
+import org.broadinstitute.sting.utils.recalibration.BaseRecalibration;
 import org.broadinstitute.sting.utils.sam.GATKSAMRecord;
 
 import java.io.PrintStream;
@@ -374,7 +375,7 @@ public class CountCovariatesWalker extends LocusWalker<CountCovariatesWalker.Cou
 
                     RecalDataManager.parseColorSpace( gatkRead );
                     gatkRead.setTemporaryAttribute( COVARS_ATTRIBUTE,
-                            RecalDataManager.computeCovariates( gatkRead, requestedCovariates ));
+                            RecalDataManager.computeCovariates( gatkRead, requestedCovariates, BaseRecalibration.BaseRecalibrationType.BASE_SUBSTITUTION ));
                 }
 
                 // Skip this position if base quality is zero

public/java/src/org/broadinstitute/sting/gatk/walkers/recalibration/Covariate.java

@@ -1,6 +1,7 @@
 package org.broadinstitute.sting.gatk.walkers.recalibration;
 
 import net.sf.samtools.SAMRecord;
+import org.broadinstitute.sting.utils.recalibration.BaseRecalibration;
 
 /*
  * Copyright (c) 2009 The Broad Institute
@@ -32,7 +33,7 @@ import net.sf.samtools.SAMRecord;
  * User: rpoplin
  * Date: Oct 30, 2009
  *
- * The Covariate interface. A Covariate is a feature used in the recalibration that can be picked out of the read, offset, and corresponding reference bases
+ * The Covariate interface. A Covariate is a feature used in the recalibration that can be picked out of the read.
  * In general most error checking and adjustments to the data are done before the call to the covariates getValue methods in order to speed up the code.
  * This unfortunately muddies the code, but most of these corrections can be done per read while the covariates get called per base, resulting in a big speed up.
  */
@@ -40,9 +41,10 @@ import net.sf.samtools.SAMRecord;
 public interface Covariate {
     public void initialize( RecalibrationArgumentCollection RAC ); // Initialize any member variables using the command-line arguments passed to the walkers
     public Comparable getValue( String str ); // Used to get the covariate's value from input csv file in TableRecalibrationWalker
-    public void getValues( SAMRecord read, Comparable[] comparable ); //Takes an array of size (at least) read.getReadLength() and fills it with covariate
+    public void getValues( SAMRecord read, Comparable[] comparable, BaseRecalibration.BaseRecalibrationType modelType );
-        //values for each position in the read. This method was created as an optimization over calling getValue( read, offset ) for each offset and allows
+            //Takes an array of size (at least) read.getReadLength() and fills it with covariate
-        //read-specific calculations to be done just once rather than for each offset.
+            //values for each position in the read. This method was created as an optimization over calling getValue( read, offset ) for each offset and allows
+            //read-specific calculations to be done just once rather than for each offset.
 }
 
 interface RequiredCovariate extends Covariate {

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

@@ -4,6 +4,7 @@ import net.sf.samtools.SAMRecord;
 import org.broadinstitute.sting.utils.BaseUtils;
 import org.broadinstitute.sting.utils.NGSPlatform;
 import org.broadinstitute.sting.utils.exceptions.UserException;
+import org.broadinstitute.sting.utils.recalibration.BaseRecalibration;
 import org.broadinstitute.sting.utils.sam.GATKSAMRecord;
 
 import java.util.EnumSet;
@@ -51,6 +52,7 @@ public class CycleCovariate implements StandardCovariate {
     private final static EnumSet<NGSPlatform> FLOW_CYCLE_PLATFORMS = EnumSet.of(NGSPlatform.LS454,  NGSPlatform.ION_TORRENT);
 
     // Initialize any member variables using the command-line arguments passed to the walkers
+    @Override
     public void initialize( final RecalibrationArgumentCollection RAC ) {
         if( RAC.DEFAULT_PLATFORM != null ) {
             if( RAC.DEFAULT_PLATFORM.equalsIgnoreCase( "SLX" ) || RAC.DEFAULT_PLATFORM.equalsIgnoreCase( "ILLUMINA" ) ||
@@ -63,7 +65,8 @@ public class CycleCovariate implements StandardCovariate {
     }
 
     // Used to pick out the covariate's value from attributes of the read
-    public void getValues(SAMRecord read, Comparable[] comparable) {
+    @Override
+    public void getValues( final SAMRecord read, final Comparable[] comparable, final BaseRecalibration.BaseRecalibrationType modelType ) {
 
         //-----------------------------
         // Illumina, Solid, PacBio, and Complete Genomics
@@ -164,6 +167,7 @@ public class CycleCovariate implements StandardCovariate {
     }
 
     // Used to get the covariate's value from input csv file in TableRecalibrationWalker
+    @Override
     public final Comparable getValue( final String str ) {
         return Integer.parseInt( str );
     }

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

@@ -2,6 +2,7 @@ package org.broadinstitute.sting.gatk.walkers.recalibration;
 
 import net.sf.samtools.SAMRecord;
 import org.broadinstitute.sting.utils.BaseUtils;
+import org.broadinstitute.sting.utils.recalibration.BaseRecalibration;
 
 import java.util.HashMap;
 
@@ -48,6 +49,7 @@ public class DinucCovariate implements StandardCovariate {
     private HashMap<Integer, Dinuc> dinucHashMap;
 
     // Initialize any member variables using the command-line arguments passed to the walkers
+    @Override
     public void initialize( final RecalibrationArgumentCollection RAC ) {
         final byte[] BASES = { (byte)'A', (byte)'C', (byte)'G', (byte)'T' };
         dinucHashMap = new HashMap<Integer, Dinuc>();
@@ -60,45 +62,11 @@ public class DinucCovariate implements StandardCovariate {
         dinucHashMap.put( Dinuc.hashBytes(NO_CALL, NO_CALL), NO_DINUC );
     }
 
-    /*
-    // Used to pick out the covariate's value from attributes of the read
-    public final Comparable getValue( final SAMRecord read, final int offset ) {
-
-        byte base;
-        byte prevBase;
-        final 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() ) {
-            // No dinuc at the beginning of the read
-            if( offset == bases.length-1 ) {
-                return NO_DINUC;
-            }
-            base = (byte)BaseUtils.simpleComplement( (char)(bases[offset]) );
-            // Note: We are using the previous base in the read, not the previous base in the reference. This is done in part to be consistent with unmapped reads.
-            prevBase = (byte)BaseUtils.simpleComplement( (char)(bases[offset + 1]) );
-        } else {
-            // No dinuc at the beginning of the read
-            if( offset == 0 ) {
-                return NO_DINUC;
-            }
-            base = bases[offset];
-            // Note: We are using the previous base in the read, not the previous base in the reference. This is done in part to be consistent with unmapped reads.
-            prevBase = bases[offset - 1];
-        }
-
-        // Make sure the previous base is good
-        if( !BaseUtils.isRegularBase( prevBase ) ) {
-            return NO_DINUC;
-        }
-
-        return dinucHashMap.get( Dinuc.hashBytes( prevBase, base ) );
-    }
-    */
-
     /**
      * Takes an array of size (at least) read.getReadLength() and fills it with the covariate values for each position in the read.
      */
-    public void getValues( SAMRecord read, Comparable[] result ) {
+    @Override
+    public void getValues( final SAMRecord read, final Comparable[] comparable, final BaseRecalibration.BaseRecalibrationType modelType ) {
         final HashMap<Integer, Dinuc> dinucHashMapRef = this.dinucHashMap; //optimize access to dinucHashMap
         final int readLength = read.getReadLength();
         final boolean negativeStrand = read.getReadNegativeStrandFlag();
@@ -111,7 +79,7 @@ public class DinucCovariate implements StandardCovariate {
         if(negativeStrand) {
             bases = BaseUtils.simpleReverseComplement(bases); //this is NOT in-place
         }
-        result[0] = NO_DINUC; // No dinuc at the beginning of the read
+        comparable[0] = NO_DINUC; // No dinuc at the beginning of the read
 
         prevBase = bases[0];
         offset++;
@@ -120,20 +88,21 @@ public class DinucCovariate implements StandardCovariate {
              // previous base in the reference. This is done in part to be consistent with unmapped reads.
              base = bases[offset];
              if( BaseUtils.isRegularBase( prevBase ) ) {
-                 result[offset] = dinucHashMapRef.get( Dinuc.hashBytes( prevBase, base ) );
+                 comparable[offset] = dinucHashMapRef.get( Dinuc.hashBytes( prevBase, base ) );
              } else {
-                 result[offset] = NO_DINUC;
+                 comparable[offset] = NO_DINUC;
              }
 
              offset++;
              prevBase = base;
         }
         if(negativeStrand) {
-            reverse( result );
+            reverse( comparable );
         }
     }
 
     // Used to get the covariate's value from input csv file in TableRecalibrationWalker
+    @Override
     public final Comparable getValue( final String str ) {
         byte[] bytes = str.getBytes();
         final Dinuc returnDinuc = dinucHashMap.get( Dinuc.hashBytes( bytes[0], bytes[1] ) );
@@ -143,7 +112,6 @@ public class DinucCovariate implements StandardCovariate {
         return returnDinuc;
     }
 
-
     /**
      * Reverses the given array in place.
      *

public/java/src/org/broadinstitute/sting/gatk/walkers/recalibration/GCContentCovariate.java

@@ -1,6 +1,7 @@
 package org.broadinstitute.sting.gatk.walkers.recalibration;
 
 import net.sf.samtools.SAMRecord;
+import org.broadinstitute.sting.utils.recalibration.BaseRecalibration;
 
 /*
  * Copyright (c) 2010 The Broad Institute
@@ -41,12 +42,13 @@ public class GCContentCovariate implements ExperimentalCovariate {
     int numBack = 7;
 
     // Initialize any member variables using the command-line arguments passed to the walkers
+    @Override
     public void initialize( final RecalibrationArgumentCollection RAC ) {
         numBack = RAC.HOMOPOLYMER_NBACK;
     }
 
     // Used to pick out the covariate's value from attributes of the read
-    public final Comparable getValue( final SAMRecord read, final int offset ) {
+    private final Comparable getValue( final SAMRecord read, final int offset ) {
 
         // ATTGCCCCGTAAAAAAAGAGAA
         // 0000123456654321001122
@@ -75,18 +77,17 @@ public class GCContentCovariate implements ExperimentalCovariate {
             return -1;
         }
     }
-    
+
-    public void getValues(SAMRecord read, Comparable[] comparable) {
+    @Override
+    public void getValues( final SAMRecord read, final Comparable[] comparable, final BaseRecalibration.BaseRecalibrationType modelType ) {
         for(int iii = 0; iii < read.getReadLength(); iii++) {
             comparable[iii] = getValue(read, iii); // BUGBUG: this can be optimized
         }
     }
 
     // Used to get the covariate's value from input csv file in TableRecalibrationWalker
+    @Override
     public final Comparable getValue( final String str ) {
         return Integer.parseInt( str );
     }
-
-
-
 }

public/java/src/org/broadinstitute/sting/gatk/walkers/recalibration/HomopolymerCovariate.java

@@ -1,6 +1,7 @@
 package org.broadinstitute.sting.gatk.walkers.recalibration;
 
 import net.sf.samtools.SAMRecord;
+import org.broadinstitute.sting.utils.recalibration.BaseRecalibration;
 
 /*
  * Copyright (c) 2009 The Broad Institute
@@ -43,12 +44,13 @@ public class HomopolymerCovariate implements ExperimentalCovariate {
     int numBack = 7;
 
     // Initialize any member variables using the command-line arguments passed to the walkers
+    @Override
     public void initialize( final RecalibrationArgumentCollection RAC ) {
         numBack = RAC.HOMOPOLYMER_NBACK;
     }
 
     // Used to pick out the covariate's value from attributes of the read
-    public final Comparable getValue( final SAMRecord read, final int offset ) {
+    private final Comparable getValue( final SAMRecord read, final int offset ) {
 
         // This block of code is for if you don't want to only count consecutive bases
         // ATTGCCCCGTAAAAAAAAATA
@@ -90,15 +92,16 @@ public class HomopolymerCovariate implements ExperimentalCovariate {
         return numAgree;
     }
 
-    public void getValues(SAMRecord read, Comparable[] comparable) {
+    @Override
+    public void getValues( final SAMRecord read, final Comparable[] comparable, final BaseRecalibration.BaseRecalibrationType modelType ) {
         for(int iii = 0; iii < read.getReadLength(); iii++) {
             comparable[iii] = getValue(read, iii); // BUGBUG: this can be optimized
         }
     }
 
     // Used to get the covariate's value from input csv file in TableRecalibrationWalker
+    @Override
     public final Comparable getValue( final String str ) {
         return Integer.parseInt( str );
     }
-
 }

public/java/src/org/broadinstitute/sting/gatk/walkers/recalibration/MappingQualityCovariate.java

@@ -1,6 +1,7 @@
 package org.broadinstitute.sting.gatk.walkers.recalibration;
 
 import net.sf.samtools.SAMRecord;
+import org.broadinstitute.sting.utils.recalibration.BaseRecalibration;
 
 /*
  * Copyright (c) 2009 The Broad Institute
@@ -38,23 +39,25 @@ import net.sf.samtools.SAMRecord;
 public class MappingQualityCovariate implements ExperimentalCovariate {
 
     // Initialize any member variables using the command-line arguments passed to the walkers
+    @Override
     public void initialize( final RecalibrationArgumentCollection RAC ) {
     }
 
     // Used to pick out the covariate's value from attributes of the read
-    public final Comparable getValue( final SAMRecord read, final int offset ) {
+    private final Comparable getValue( final SAMRecord read, final int offset ) {
         return read.getMappingQuality();
     }
 
     // Used to get the covariate's value from input csv file in TableRecalibrationWalker
+    @Override
     public final Comparable getValue( final String str ) {
         return Integer.parseInt( str );
     }
 
-    public void getValues(SAMRecord read, Comparable[] comparable) {
+    @Override
+    public void getValues( final SAMRecord read, final Comparable[] comparable, final BaseRecalibration.BaseRecalibrationType modelType ) {
         for(int iii = 0; iii < read.getReadLength(); iii++) {
             comparable[iii] = getValue(read, iii); // BUGBUG: this can be optimized
         }
     }
-
 }

public/java/src/org/broadinstitute/sting/gatk/walkers/recalibration/MinimumNQSCovariate.java

@@ -1,6 +1,7 @@
 package org.broadinstitute.sting.gatk.walkers.recalibration;
 
 import net.sf.samtools.SAMRecord;
+import org.broadinstitute.sting.utils.recalibration.BaseRecalibration;
 
 /*
  * Copyright (c) 2009 The Broad Institute
@@ -41,12 +42,13 @@ public class MinimumNQSCovariate implements ExperimentalCovariate {
     private int windowReach; // How far in each direction from the current base to look
 
     // Initialize any member variables using the command-line arguments passed to the walkers
+    @Override
     public void initialize( final RecalibrationArgumentCollection RAC ) {
         windowReach = RAC.WINDOW_SIZE / 2; // integer division
     }
 
     // Used to pick out the covariate's value from attributes of the read
-    public final Comparable getValue( final SAMRecord read, final int offset ) {
+    private final Comparable getValue( final SAMRecord read, final int offset ) {
 
         // Loop over the list of base quality scores in the window and find the minimum
         final byte[] quals = read.getBaseQualities();
@@ -61,14 +63,16 @@ public class MinimumNQSCovariate implements ExperimentalCovariate {
         return minQual;
     }
 
-    // Used to get the covariate's value from input csv file in TableRecalibrationWalker
+    @Override
-    public final Comparable getValue( final String str ) {
+    public void getValues( final SAMRecord read, final Comparable[] comparable, final BaseRecalibration.BaseRecalibrationType modelType ) {
-        return Integer.parseInt( str );
-    }
-
-    public void getValues(SAMRecord read, Comparable[] comparable) {
         for(int iii = 0; iii < read.getReadLength(); iii++) {
             comparable[iii] = getValue(read, iii); // BUGBUG: this can be optimized
         }
     }
+
+    // Used to get the covariate's value from input csv file in TableRecalibrationWalker
+    @Override
+    public final Comparable getValue( final String str ) {
+        return Integer.parseInt( str );
+    }
 }

public/java/src/org/broadinstitute/sting/gatk/walkers/recalibration/PositionCovariate.java

@@ -1,6 +1,7 @@
 package org.broadinstitute.sting.gatk.walkers.recalibration;
 
 import net.sf.samtools.SAMRecord;
+import org.broadinstitute.sting.utils.recalibration.BaseRecalibration;
 
 /*
  * Copyright (c) 2009 The Broad Institute
@@ -39,11 +40,12 @@ import net.sf.samtools.SAMRecord;
 public class PositionCovariate implements ExperimentalCovariate {
 
     // Initialize any member variables using the command-line arguments passed to the walkers
+    @Override
     public void initialize( final RecalibrationArgumentCollection RAC ) {
     }
 
     // Used to pick out the covariate's value from attributes of the read
-    public final Comparable getValue( final SAMRecord read, final int offset ) {
+    private final Comparable getValue( final SAMRecord read, final int offset ) {
         int cycle = offset;
         if( read.getReadNegativeStrandFlag() ) {
             cycle = read.getReadLength() - (offset + 1);
@@ -51,15 +53,16 @@ public class PositionCovariate implements ExperimentalCovariate {
         return cycle;
     }
 
-    // Used to get the covariate's value from input csv file in TableRecalibrationWalker
+    @Override
-    public final Comparable getValue( final String str ) {
+    public void getValues( final SAMRecord read, final Comparable[] comparable, final BaseRecalibration.BaseRecalibrationType modelType ) {
-        return Integer.parseInt( str );
-    }
-
-    public void getValues(SAMRecord read, Comparable[] comparable) {
         for(int iii = 0; iii < read.getReadLength(); iii++) {
             comparable[iii] = getValue(read, iii); // BUGBUG: this can be optimized
         }
     }
 
+    // Used to get the covariate's value from input csv file in TableRecalibrationWalker
+    @Override
+    public final Comparable getValue( final String str ) {
+        return Integer.parseInt( str );
+    }
 }

public/java/src/org/broadinstitute/sting/gatk/walkers/recalibration/PrimerRoundCovariate.java

@@ -1,6 +1,7 @@
 package org.broadinstitute.sting.gatk.walkers.recalibration;
 
 import net.sf.samtools.SAMRecord;
+import org.broadinstitute.sting.utils.recalibration.BaseRecalibration;
 
 /*
  * Copyright (c) 2009 The Broad Institute
@@ -40,11 +41,12 @@ import net.sf.samtools.SAMRecord;
 public class PrimerRoundCovariate implements ExperimentalCovariate {
 
     // Initialize any member variables using the command-line arguments passed to the walkers
+    @Override
     public void initialize( final RecalibrationArgumentCollection RAC ) {
     }
 
     // Used to pick out the covariate's value from attributes of the read
-    public final Comparable getValue( final SAMRecord read, final int offset ) {
+    private final Comparable getValue( final SAMRecord read, final int offset ) {
         if( read.getReadGroup().getPlatform().equalsIgnoreCase( "SOLID" ) || read.getReadGroup().getPlatform().equalsIgnoreCase( "ABI_SOLID" ) ) {
             int pos = offset;
             if( read.getReadNegativeStrandFlag() ) {
@@ -57,14 +59,16 @@ public class PrimerRoundCovariate implements ExperimentalCovariate {
 
     }
 
-    // Used to get the covariate's value from input csv file in TableRecalibrationWalker
+    @Override
-    public final Comparable getValue( final String str ) {
+    public void getValues( final SAMRecord read, final Comparable[] comparable, final BaseRecalibration.BaseRecalibrationType modelType ) {
-        return Integer.parseInt( str );
-    }
-
-    public void getValues(SAMRecord read, Comparable[] comparable) {
         for(int iii = 0; iii < read.getReadLength(); iii++) {
             comparable[iii] = getValue(read, iii); // BUGBUG: this can be optimized
         }
     }
+
+    // Used to get the covariate's value from input csv file in TableRecalibrationWalker
+    @Override
+    public final Comparable getValue( final String str ) {
+        return Integer.parseInt( str );
+    }
 }

public/java/src/org/broadinstitute/sting/gatk/walkers/recalibration/QualityScoreCovariate.java

@@ -1,6 +1,9 @@
 package org.broadinstitute.sting.gatk.walkers.recalibration;
 
 import net.sf.samtools.SAMRecord;
+import org.broadinstitute.sting.utils.recalibration.BaseRecalibration;
+
+import java.util.Arrays;
 
 /*
  * Copyright (c) 2009 The Broad Institute
@@ -38,26 +41,26 @@ import net.sf.samtools.SAMRecord;
 public class QualityScoreCovariate implements RequiredCovariate {
 
     // Initialize any member variables using the command-line arguments passed to the walkers
+    @Override
     public void initialize( final RecalibrationArgumentCollection RAC ) {
     }
 
-    /*
+    @Override
-    // Used to pick out the covariate's value from attributes of the read
+    public void getValues( final SAMRecord read, final Comparable[] comparable, final BaseRecalibration.BaseRecalibrationType modelType ) {
-    public final Comparable getValue( final SAMRecord read, final int offset ) {
+        if( modelType == BaseRecalibration.BaseRecalibrationType.BASE_SUBSTITUTION ) {
-        return (int)(read.getBaseQualities()[offset]);
+            byte[] baseQualities = read.getBaseQualities();
-    }
+            for(int i = 0; i < read.getReadLength(); i++) {
-    */
+                comparable[i] = (int) baseQualities[i];
-
+            }
-    public void getValues(SAMRecord read, Comparable[] comparable) {
+        } else { // model == BASE_INSERTION || model == BASE_DELETION
-        byte[] baseQualities = read.getBaseQualities();
+            Arrays.fill(comparable, 45); // Some day in the future when base insertion and base deletion quals exist the samtools API will
-        for(int i = 0; i < read.getReadLength(); i++) {
+                                         // be updated and the original quals will be pulled here, but for now we assume the original quality is a flat Q45
-            comparable[i] = (int) baseQualities[i];
         }
     }
 
     // Used to get the covariate's value from input csv file in TableRecalibrationWalker
+    @Override
     public final Comparable getValue( final String str ) {
         return Integer.parseInt( str );
     }
-
 }

public/java/src/org/broadinstitute/sting/gatk/walkers/recalibration/ReadGroupCovariate.java

@@ -1,6 +1,7 @@
 package org.broadinstitute.sting.gatk.walkers.recalibration;
 
 import net.sf.samtools.SAMRecord;
+import org.broadinstitute.sting.utils.recalibration.BaseRecalibration;
 
 /*
  * Copyright (c) 2009 The Broad Institute
@@ -35,22 +36,17 @@ import net.sf.samtools.SAMRecord;
  * The Read Group covariate.
  */
 
-public class ReadGroupCovariate implements RequiredCovariate{
+public class ReadGroupCovariate implements RequiredCovariate {
 
     public static final String defaultReadGroup = "DefaultReadGroup";
 
     // Initialize any member variables using the command-line arguments passed to the walkers
+    @Override
     public void initialize( final RecalibrationArgumentCollection RAC ) {
     }
 
-    /*
+    @Override
-    // Used to pick out the covariate's value from attributes of the read
+    public void getValues( final SAMRecord read, final Comparable[] comparable, final BaseRecalibration.BaseRecalibrationType modelType ) {
-    public final Comparable getValue( final SAMRecord read, final int offset ) {
-        return read.getReadGroup().getReadGroupId();
-    }
-    */
-
-    public void getValues(SAMRecord read, Comparable[] comparable) {
         final String readGroupId = read.getReadGroup().getReadGroupId();
         for(int i = 0; i < read.getReadLength(); i++) {
             comparable[i] = readGroupId;
@@ -58,10 +54,10 @@ public class ReadGroupCovariate implements RequiredCovariate{
     }
 
     // Used to get the covariate's value from input csv file in TableRecalibrationWalker
+    @Override
     public final Comparable getValue( final String str ) {
         return str;
     }
-
 }
 
 

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

@@ -33,6 +33,7 @@ import org.broadinstitute.sting.utils.Utils;
 import org.broadinstitute.sting.utils.collections.NestedHashMap;
 import org.broadinstitute.sting.utils.exceptions.ReviewedStingException;
 import org.broadinstitute.sting.utils.exceptions.UserException;
+import org.broadinstitute.sting.utils.recalibration.BaseRecalibration;
 import org.broadinstitute.sting.utils.sam.AlignmentUtils;
 import org.broadinstitute.sting.utils.sam.GATKSAMReadGroupRecord;
 import org.broadinstitute.sting.utils.sam.GATKSAMRecord;
@@ -571,7 +572,7 @@ public class RecalDataManager {
      * value for the ith position in the read and the jth covariate in
      * reqeustedCovariates list.
      */
-     public static Comparable[][] computeCovariates(final GATKSAMRecord gatkRead, final List<Covariate> requestedCovariates) {
+     public static Comparable[][] computeCovariates( final GATKSAMRecord gatkRead, final List<Covariate> requestedCovariates, final BaseRecalibration.BaseRecalibrationType modelType ) {
          //compute all covariates for this read
          final List<Covariate> requestedCovariatesRef = requestedCovariates;
          final int numRequestedCovariates = requestedCovariatesRef.size();
@@ -582,7 +583,7 @@ public class RecalDataManager {
 
          // Loop through the list of requested covariates and compute the values of each covariate for all positions in this read
          for( int i = 0; i < numRequestedCovariates; i++ ) {
-             requestedCovariatesRef.get(i).getValues( gatkRead, tempCovariateValuesHolder );
+             requestedCovariatesRef.get(i).getValues( gatkRead, tempCovariateValuesHolder, modelType );
              for(int j = 0; j < readLength; j++) {
                  //copy values into a 2D array that allows all covar types to be extracted at once for
                  //an offset j by doing covariateValues_offset_x_covar[j]. This avoids the need to later iterate over covar types.

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

@@ -39,6 +39,7 @@ import org.broadinstitute.sting.utils.classloader.PluginManager;
 import org.broadinstitute.sting.utils.collections.NestedHashMap;
 import org.broadinstitute.sting.utils.exceptions.DynamicClassResolutionException;
 import org.broadinstitute.sting.utils.exceptions.UserException;
+import org.broadinstitute.sting.utils.recalibration.BaseRecalibration;
 import org.broadinstitute.sting.utils.sam.GATKSAMRecord;
 import org.broadinstitute.sting.utils.text.TextFormattingUtils;
 import org.broadinstitute.sting.utils.text.XReadLines;
@@ -398,7 +399,7 @@ public class TableRecalibrationWalker extends ReadWalker<SAMRecord, SAMFileWrite
 
         //compute all covariate values for this read
         final Comparable[][] covariateValues_offset_x_covar =
-            RecalDataManager.computeCovariates(read, requestedCovariates);
+            RecalDataManager.computeCovariates(read, requestedCovariates, BaseRecalibration.BaseRecalibrationType.BASE_SUBSTITUTION);
 
         // For each base in the read
         for( int offset = 0; offset < read.getReadLength(); offset++ ) {

public/java/src/org/broadinstitute/sting/gatk/walkers/validation/validationsiteselector/KeepAFSpectrumFrequencySelector.java

@@ -71,25 +71,24 @@ public class KeepAFSpectrumFrequencySelector extends FrequencyModeSelector {
             // recompute AF,AC,AN based on genotypes:
             // todo - - maybe too inefficient??
             VariantContextUtils.calculateChromosomeCounts(vc, attributes, false);
-            afArray =  new double[] {Double.valueOf((String)attributes.get(VCFConstants.ALLELE_FREQUENCY_KEY))};
+        }
-        }   else {
-            // sites-only vc or we explicitly tell to ignore genotypes; we trust the AF field if present
-            if ( vc.hasAttribute(VCFConstants.ALLELE_FREQUENCY_KEY) )  {
-                String afo = vc.getAttributeAsString(VCFConstants.ALLELE_FREQUENCY_KEY, null);
 
-                if (afo.contains(",")) {
+        // sites-only vc or we explicitly tell to ignore genotypes; we trust the AF field if present
-                    String[] afs = afo.split(",");
+        if ( vc.hasAttribute(VCFConstants.ALLELE_FREQUENCY_KEY) )  {
-                    afs[0] = afs[0].substring(1,afs[0].length());
+            String afo = vc.getAttributeAsString(VCFConstants.ALLELE_FREQUENCY_KEY, null);
-                    afs[afs.length-1] = afs[afs.length-1].substring(0,afs[afs.length-1].length()-1);
 
-                    afArray = new double[afs.length];
+            if (afo.contains(",")) {
+                String[] afs = afo.split(",");
+                afs[0] = afs[0].substring(1,afs[0].length());
+                afs[afs.length-1] = afs[afs.length-1].substring(0,afs[afs.length-1].length()-1);
 
-                    for (int k=0; k < afArray.length; k++)
+                afArray = new double[afs.length];
-                        afArray[k] = Double.valueOf(afs[k]);
+
-                }
+                for (int k=0; k < afArray.length; k++)
-                else
+                    afArray[k] = Double.valueOf(afs[k]);
-                    afArray = new double[] {Double.valueOf(afo)};
             }
+            else
+                afArray = new double[] {Double.valueOf(afo)};
         }
 
 

public/java/src/org/broadinstitute/sting/utils/recalibration/BaseRecalibration.java

@@ -177,13 +177,13 @@ public class BaseRecalibration {
         dataManager.addToAllTables( key, datum, QualityUtils.MIN_USABLE_Q_SCORE ); //BUGBUG: used to be Q5 now is Q6, probably doesn't matter
     }
     
-    public byte[] recalibrateRead( final GATKSAMRecord read, final byte[] originalQuals ) {
+    public byte[] recalibrateRead( final GATKSAMRecord read, final byte[] originalQuals, final BaseRecalibrationType modelType ) {
 
         final byte[] recalQuals = originalQuals.clone();
         
         //compute all covariate values for this read
         final Comparable[][] covariateValues_offset_x_covar =
-                RecalDataManager.computeCovariates(read, requestedCovariates);
+                RecalDataManager.computeCovariates(read, requestedCovariates, modelType);
     
         // For each base in the read
         for( int offset = 0; offset < read.getReadLength(); offset++ ) {

public/java/src/org/broadinstitute/sting/utils/sam/GATKSAMRecord.java

@@ -27,6 +27,7 @@ package org.broadinstitute.sting.utils.sam;
 import net.sf.samtools.*;
 import org.broadinstitute.sting.gatk.GenomeAnalysisEngine;
 import org.broadinstitute.sting.utils.NGSPlatform;
+import org.broadinstitute.sting.utils.recalibration.BaseRecalibration;
 
 import java.util.Arrays;
 import java.util.HashMap;
@@ -162,9 +163,11 @@ public class GATKSAMRecord extends BAMRecord {
         return super.equals(o);
     }
 
-    /*
+
     @Override
     public byte[] getBaseQualities() {
+        return super.getBaseQualities();
+        /*
         if( getAttribute( BQSR_BASES_HAVE_BEEN_RECALIBRATED_TAG ) != null ) {
             return super.getBaseQualities();
         } else {
@@ -178,8 +181,8 @@ public class GATKSAMRecord extends BAMRecord {
                 return super.getBaseQualities();
             }
         }
+        */
     }
-    */
 
     /**
      * Accessors for base insertion and base deletion quality scores
@@ -192,7 +195,7 @@ public class GATKSAMRecord extends BAMRecord {
             // if the recal data was populated in the engine then recalibrate the quality scores on the fly
             // else give default values which are flat Q45
             if( GenomeAnalysisEngine.hasBaseRecalibration() ) {
-                quals = GenomeAnalysisEngine.getBaseRecalibration().recalibrateRead( this, quals ); // the original quals here are the flat base insertion/deletion quals, NOT the original base qualities
+                quals = GenomeAnalysisEngine.getBaseRecalibration().recalibrateRead( this, quals, BaseRecalibration.BaseRecalibrationType.BASE_INSERTION ); // the original quals here are the flat base insertion/deletion quals, NOT the original base qualities
             }
             // add the qual array to the read so that we don't have to do the recalibration work again
             setAttribute( BQSR_BASE_INSERTION_QUALITIES, quals );
@@ -208,7 +211,7 @@ public class GATKSAMRecord extends BAMRecord {
             // if the recal data was populated in the engine then recalibrate the quality scores on the fly
             // else give default values which are flat Q45
             if( GenomeAnalysisEngine.hasBaseRecalibration() ) {
-                quals = GenomeAnalysisEngine.getBaseRecalibration().recalibrateRead( this, quals ); // the original quals here are the flat base insertion/deletion quals, NOT the original base qualities
+                quals = GenomeAnalysisEngine.getBaseRecalibration().recalibrateRead( this, quals, BaseRecalibration.BaseRecalibrationType.BASE_DELETION ); // the original quals here are the flat base insertion/deletion quals, NOT the original base qualities
             }
             // add the qual array to the read so that we don't have to do the recalibration work again
             setAttribute( BQSR_BASE_DELETION_QUALITIES, quals );