more cleanup

git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@1631 348d0f76-0448-11de-a6fe-93d51630548a

java/src/org/broadinstitute/sting/playground/gatk/walkers/varianteval/CallableBasesAnalysis.java

@@ -3,10 +3,9 @@ package org.broadinstitute.sting.playground.gatk.walkers.varianteval;
 import org.broadinstitute.sting.gatk.contexts.AlignmentContext;
 import org.broadinstitute.sting.gatk.refdata.RefMetaDataTracker;
 import org.broadinstitute.sting.utils.genotype.DiploidGenotype;
+import org.broadinstitute.sting.utils.genotype.Genotype;
 import org.broadinstitute.sting.utils.genotype.VariantBackedByGenotype;
 import org.broadinstitute.sting.utils.genotype.Variation;
-import org.broadinstitute.sting.utils.genotype.Genotype;
-import org.broadinstitute.sting.utils.Utils;
 
 import java.util.ArrayList;
 import java.util.List;
@@ -23,7 +22,6 @@ import java.util.List;
 public class CallableBasesAnalysis extends BasicVariantAnalysis implements GenotypeAnalysis {
     long all_bases = 0;
     long all_calls = 0;
-    //final static double[] Qthresholds = { 10, 20, 30, 40, 50, 100, 200, 500, 1000 };
     final static double[] thresholds = {0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 15, 20, 50, 100};
     long[] discoverable_bases = new long[thresholds.length];
     long[] genotypable_bases = new long[thresholds.length];
@@ -64,14 +62,14 @@ public class CallableBasesAnalysis extends BasicVariantAnalysis implements Genot
 
         // we actually have a record here
         if (!(eval instanceof VariantBackedByGenotype)) {            // evaluation record isn't a genotype, die!
-            throw new RuntimeException("Evaluation track isn't an Genotype!");
+            throw new RuntimeException("Evaluation track isn't backed by a Genotype!");
         }
 
         all_calls++;
         // For every threshold, updated discoverable and callable
         for (int i = 0; i < thresholds.length; i++) {
             double threshold = thresholds[i];
-            DiploidGenotype g = DiploidGenotype.valueOf(Utils.dupString(ref, 2));
+            DiploidGenotype g = DiploidGenotype.createHomGenotype(ref);
             Genotype genotype = ((VariantBackedByGenotype) eval).getGenotype(g);
             // update discoverable
             if (eval.isSNP() && eval.getNegLog10PError() >= threshold)

java/src/org/broadinstitute/sting/playground/gatk/walkers/varianteval/GenotypeConcordance.java

@@ -2,9 +2,8 @@ package org.broadinstitute.sting.playground.gatk.walkers.varianteval;
 
 import org.broadinstitute.sting.gatk.contexts.AlignmentContext;
 import org.broadinstitute.sting.gatk.refdata.RefMetaDataTracker;
-import org.broadinstitute.sting.utils.genotype.DiploidGenotype;
 import org.broadinstitute.sting.utils.StingException;
-import org.broadinstitute.sting.utils.Utils;
+import org.broadinstitute.sting.utils.genotype.DiploidGenotype;
 import org.broadinstitute.sting.utils.genotype.VariantBackedByGenotype;
 import org.broadinstitute.sting.utils.genotype.Variation;
 
@@ -50,7 +49,7 @@ public class GenotypeConcordance extends BasicVariantAnalysis implements Genotyp
         if ((chip != null && !(chip instanceof VariantBackedByGenotype) || (eval != null && !(eval instanceof VariantBackedByGenotype))))
             throw new StingException("Failure: trying to analyze genotypes of non-genotype data");
 
-        DiploidGenotype g = DiploidGenotype.valueOf(Utils.dupString(ref, 2));
+        DiploidGenotype g = DiploidGenotype.createHomGenotype(ref);
         int truthIndex, callIndex;
         if (chip == null)
             truthIndex = UNKNOWN;

java/src/org/broadinstitute/sting/utils/genotype/BasicGenotype.java

@@ -4,26 +4,38 @@ import org.broadinstitute.sting.utils.GenomeLoc;
 
 
 /**
- * 
+ * @author aaron
- * @author aaron 
+ *         <p/>
- * 
+ *         Class BasicGenotype
- * Class BasicGenotype
+ *         <p/>
- *
+ *         represents a basic genotype object.  That means that is
- * represents a basic genotype object
+ *         an implementation for a basic genotype call, given the genotype
+ *         string, the ref base, the confidence score, and the location. This
+ *         class currently only represents point genotypes, not indels
  */
 public class BasicGenotype implements Genotype {
     // the genotype string
     private String mGenotype;
+
+    // our location
     private GenomeLoc mLocation;
+
+    // the reference base.
     private char mRef;
-    private double mNetLog10PError;
+
+    // the confidence score
+    private double mNegLog10PError;
 
     /**
-     * create a basic genotype
+     * create a basic genotype, given the following fields
-     * @param genotype
+     *
+     * @param location       the genomic location
+     * @param genotype       the genotype, as a string, where ploidy = string.length
+     * @param ref            the reference base as a char
+     * @param negLog10PError the confidence score
      */
-    public BasicGenotype(GenomeLoc location, String genotype, char ref, double netLog10PError) {
+    public BasicGenotype(GenomeLoc location, String genotype, char ref, double negLog10PError) {
-        mNetLog10PError = netLog10PError;
+        mNegLog10PError = negLog10PError;
         mGenotype = genotype;
         mLocation = location;
         mRef = ref;
@@ -32,15 +44,15 @@ public class BasicGenotype implements Genotype {
     /**
      * get the -1 * (log 10 of the error value)
      *
-     * @return the log based error estimate
+     * @return the negitive log based error estimate
      */
     @Override
     public double getNegLog10PError() {
-        return mNetLog10PError;
+        return mNegLog10PError;
     }
 
     /**
-     * get the bases that represent this
+     * get the bases that represent this genotype
      *
      * @return the bases, as a string
      */
@@ -60,7 +72,7 @@ public class BasicGenotype implements Genotype {
     }
 
     /**
-     * Returns true if both observed alleles are the same (regardless of whether they are ref or alt)
+     * Returns true if both observed allele bases are the same (regardless of whether they are ref or alt)
      *
      * @return true if we're homozygous, false otherwise
      */
@@ -79,7 +91,7 @@ public class BasicGenotype implements Genotype {
     }
 
     /**
-     * Returns true if observed alleles differ (regardless of whether they are ref or alt)
+     * Returns true if observed allele bases differ (regardless of whether they are ref or alt)
      *
      * @return true if we're het, false otherwise
      */
@@ -138,6 +150,6 @@ public class BasicGenotype implements Genotype {
     @Override
     public Variation toVariation() {
         if (!isVariant(this.mRef)) throw new IllegalStateException("this genotype is not a variant");
-        return new BasicVariation(this.getBases(),String.valueOf(mRef),this.getBases().length(),mLocation,mNetLog10PError);
+        return new BasicVariation(this.getBases(), String.valueOf(mRef), this.getBases().length(), mLocation, mNegLog10PError);
     }
 }

java/src/org/broadinstitute/sting/utils/genotype/BasicVariation.java

@@ -3,23 +3,31 @@ package org.broadinstitute.sting.utils.genotype;
 import org.broadinstitute.sting.utils.GenomeLoc;
 
 /**
- * Created by IntelliJ IDEA.
+ * User: aaron
- * User: aaronmckenna
  * Date: Sep 9, 2009
  * Time: 9:32:34 PM
  * <p/>
- * a basic implementation of variant
+ * a basic implementation of the Variation interface.
  */
 public class BasicVariation implements Variation {
 
+    // the bases that make up this variant
     protected final String mBases;
+
+    // the reference base
     protected final String mRef;
+
+    // the length of the event, 0 for a SNP, negitive for deletions, positive for insertions
     protected final int mLength;
+
+    // the location on the genome of the event
     protected final GenomeLoc mLocation;
+
+    // our confidence in this event, and a -(log10(Error))
     protected final double mConfidence;
 
     /**
-     * the constructor
+     * create a basic variation, given the following parameters:
      *
      * @param bases     the bases that this variant represents
      * @param reference the reference bases
@@ -34,11 +42,21 @@ public class BasicVariation implements Variation {
         mConfidence = confidence;
     }
 
+    /**
+     * we don't know the minor allele freq. is this implementation
+     *
+     * @return -1.0.  If the freq is less than zero it means we don't know
+     */
     @Override
     public double getNonRefAlleleFrequency() {
         return -1.0;
     }
 
+    /**
+     * get the type of variation we are
+     *
+     * @return VARIANT_TYPE
+     */
     @Override
     public VARIANT_TYPE getType() {
         if (mLength > 0) return VARIANT_TYPE.INDEL;
@@ -84,7 +102,7 @@ public class BasicVariation implements Variation {
 
     @Override
     public boolean isReference() {
-        if (mLength != 0) return true;
+        if (mLength != 0) return false;
         int refIndex = 0;
         for (char c : mBases.toCharArray()) {
             if (mRef.charAt(refIndex) != c) return false;
@@ -93,7 +111,7 @@ public class BasicVariation implements Variation {
     }
 
     /**
-     * are we an insertion or a deletion? yes, then return true.  No? Well, false it is.
+     * are we an insertion or a deletion? yes, then return true.
      *
      * @return true if we're an insertion or deletion
      */
@@ -111,6 +129,8 @@ public class BasicVariation implements Variation {
     @Override
     public char getAlternativeBaseForSNP() {
         if (!this.isSNP()) throw new IllegalStateException("we're not a SNP");
+
+        // we know that if we're a snp, the reference is a single base, so charAt(0) is safe
         if (getAlternateBases().charAt(0) == this.getReference().charAt(0))
             return getAlternateBases().charAt(1);
         return getAlternateBases().charAt(0);
@@ -124,6 +144,8 @@ public class BasicVariation implements Variation {
     @Override
     public char getReferenceForSNP() {
         if (!this.isSNP()) throw new IllegalStateException("we're not a SNP");
+
+        // we know that if we're a snp, the reference is a single base, so charAt(0) is safe
         if (getAlternateBases().charAt(0) == this.getReference().charAt(0))
             return getAlternateBases().charAt(0);
         return getAlternateBases().charAt(1);

java/src/org/broadinstitute/sting/utils/genotype/DiploidGenotype.java

@@ -58,7 +58,7 @@ public enum DiploidGenotype {
      * @param hom the character to turn into a hom genotype, i.e. if it is A, then returned will be AA
      * @return the diploid genotype
      */
-    public static DiploidGenotype createGenotype(char hom) {
+    public static DiploidGenotype createHomGenotype(char hom) {
         return DiploidGenotype.valueOf((String.valueOf(hom) + String.valueOf(hom)).toUpperCase());     
     }
 }

java/test/org/broadinstitute/sting/utils/genotype/DiploidGenotypeTest.java

@@ -83,19 +83,19 @@ public class DiploidGenotypeTest extends BaseTest {
     @Test
       public void testCreateGenotype() {
         char ref = 'A';
-        DiploidGenotype g = DiploidGenotype.createGenotype(ref);
+        DiploidGenotype g = DiploidGenotype.createHomGenotype(ref);
         Assert.assertTrue("AA".equals(g.toString()));
 
         ref = 'a';
-        g = DiploidGenotype.createGenotype(ref);
+        g = DiploidGenotype.createHomGenotype(ref);
         Assert.assertTrue("AA".equals(g.toString()));
 
         ref = 't';
-        g = DiploidGenotype.createGenotype(ref);
+        g = DiploidGenotype.createHomGenotype(ref);
         Assert.assertTrue("TT".equals(g.toString()));
 
         ref = 'T';
-        g = DiploidGenotype.createGenotype(ref);
+        g = DiploidGenotype.createHomGenotype(ref);
         Assert.assertTrue("TT".equals(g.toString()));
 
     }