From dc9d40eb9adace272b395df462882968f9c356b7 Mon Sep 17 00:00:00 2001
From: depristo <depristo@348d0f76-0448-11de-a6fe-93d51630548a>
Date: Fri, 28 Aug 2009 00:19:23 +0000
Subject: [PATCH] Now requires a minimum genotype LOD before applying tests

git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@1471 348d0f76-0448-11de-a6fe-93d51630548a
---
 .../walkers/variants/VECAlleleBalance.java    | 29 ++++++++++---------
 .../variants/VECOnOffGenotypeRatio.java       | 15 +++-------
 2 files changed, 20 insertions(+), 24 deletions(-)

diff --git a/java/src/org/broadinstitute/sting/playground/gatk/walkers/variants/VECAlleleBalance.java b/java/src/org/broadinstitute/sting/playground/gatk/walkers/variants/VECAlleleBalance.java
index 09b251fbc..b00c75f45 100755
--- a/java/src/org/broadinstitute/sting/playground/gatk/walkers/variants/VECAlleleBalance.java
+++ b/java/src/org/broadinstitute/sting/playground/gatk/walkers/variants/VECAlleleBalance.java
@@ -8,9 +8,11 @@ import org.broadinstitute.sting.utils.Pair;
 import org.broadinstitute.sting.utils.GenotypeUtils;
 
 public class VECAlleleBalance implements VariantExclusionCriterion {  //extends RatioFilter {
-    //final private static GenotypeFeatureData.Tail tail = GenotypeFeatureData.Tail.TwoTailed;
+    private static final double defaultMinGenotypeConfidenceToTest = 5.0;  // TODO -- must be replaced with true confidence scoore, right now assumes LOD
+
     private boolean exclude;
     private double lowThreshold = 0.1, highThreshold = 0.85;
+    private double minGenotypeConfidenceToTest = defaultMinGenotypeConfidenceToTest;
     private double ratio;
 
     public void initialize(String arguments) {
@@ -18,20 +20,11 @@ public class VECAlleleBalance implements VariantExclusionCriterion {  //extends
             String[] argPieces = arguments.split(",");
             lowThreshold = Double.valueOf(argPieces[0]);
             highThreshold = Double.valueOf(argPieces[1]);
+            if ( argPieces.length > 2 )
+                minGenotypeConfidenceToTest = Double.valueOf(argPieces[2]);
         }
     }
 
-    /**
-     * We can only filter out het calls with the AlleleBalance filter
-     *
-     * @param variant
-     * @return
-     */
-    protected boolean applyToVariant(rodVariants variant) {
-        String genotype = variant.getBestGenotype();
-        return genotype.charAt(0) != genotype.charAt(1);
-    }
-
     /**
      * Return the count of bases matching the major (first) and minor (second) alleles as a pair.
      *
@@ -64,7 +57,17 @@ public class VECAlleleBalance implements VariantExclusionCriterion {  //extends
         Pair<Integer, Integer> counts = scoreVariant(ref, pileup, variant);
         int n = counts.first + counts.second;
         ratio = counts.first.doubleValue() / (double)n;
-        exclude = GenotypeUtils.isHet(variant) && (ratio < lowThreshold || ratio > highThreshold);
+
+        boolean highGenotypeConfidence = variant.getConsensusConfidence() > minGenotypeConfidenceToTest;
+        boolean failsHetExpectation = GenotypeUtils.isHet(variant) && (ratio < lowThreshold || ratio > highThreshold);
+        exclude = failsHetExpectation && highGenotypeConfidence;
+
+//        if ( failsHetExpectation ) {
+//            String header = highGenotypeConfidence ? "FILTER-HIGH-CONFIDENCE" : "PASS-LOW-CONFIDENCE";
+//            System.out.printf("[%s] %s getConsensusConfidence() = %f > minGenotypeConfidenceToTest = %f exclude=%b  %s%n",
+//                    header, variant.getBestGenotype(), variant.getConsensusConfidence(), minGenotypeConfidenceToTest, exclude,
+//                    ! highGenotypeConfidence ? variant : "" );
+//        }
     }
 
     public boolean useZeroQualityReads() { return false; }
diff --git a/java/src/org/broadinstitute/sting/playground/gatk/walkers/variants/VECOnOffGenotypeRatio.java b/java/src/org/broadinstitute/sting/playground/gatk/walkers/variants/VECOnOffGenotypeRatio.java
index 153e7dc3b..25056e944 100755
--- a/java/src/org/broadinstitute/sting/playground/gatk/walkers/variants/VECOnOffGenotypeRatio.java
+++ b/java/src/org/broadinstitute/sting/playground/gatk/walkers/variants/VECOnOffGenotypeRatio.java
@@ -5,6 +5,8 @@ import org.broadinstitute.sting.gatk.contexts.AlignmentContext;
 import org.broadinstitute.sting.utils.*;
 
 public class VECOnOffGenotypeRatio implements VariantExclusionCriterion { // extends RatioFilter {
+    private static final double minGenotypeConfidenceToTest = 5.0;  // TODO -- must be replaced with true confidence scoore, right now assumes LOD
+
     //final private static GenotypeFeatureData.Tail tail = GenotypeFeatureData.Tail.LeftTailed;
     private boolean exclude;
     private double threshold = 0.0;
@@ -16,16 +18,6 @@ public class VECOnOffGenotypeRatio implements VariantExclusionCriterion { // ext
         }
     }
 
-    /**
-     * On/off genotype filters can be applied to any genotype
-     *
-     * @param variant
-     * @return
-     */
-    protected boolean applyToVariant(rodVariants variant) {
-        return true;
-    }
-
     /**
      * Return the counts of bases that are on (matching the bestGenotype) and off (not matching the
      * best genotype).  On are in the first field, off in the second.
@@ -61,7 +53,8 @@ public class VECOnOffGenotypeRatio implements VariantExclusionCriterion { // ext
         Pair<Integer, Integer> counts = scoreVariant(ref, pileup, variant);
         int n = counts.first + counts.second;
         ratio = counts.first.doubleValue() / (double)n;
-        exclude = ratio < threshold;
+        boolean highGenotypeConfidence = variant.getConsensusConfidence() > minGenotypeConfidenceToTest;
+        exclude = ratio < threshold && highGenotypeConfidence;
     }
 
     public double inclusionProbability() {