Removing the AllelicVariant interface, which is replaced by the Variation interface.

git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@1770 348d0f76-0448-11de-a6fe-93d51630548a

java/src/org/broadinstitute/sting/gatk/refdata/AllelicVariant.java

@@ -1,148 +0,0 @@
-package org.broadinstitute.sting.gatk.refdata;
-
-import org.broadinstitute.sting.gatk.refdata.ReferenceOrderedDatum;
-import org.broadinstitute.sting.utils.GenomeLoc;
-
-import java.util.List;
-
-/**
- * Created by IntelliJ IDEA.
- * User: asivache
- * Date: Apr 1, 2009
- * Time: 11:02:03 AM
- * To change this template use File | Settings | File Templates.
- */
-public interface AllelicVariant extends ReferenceOrderedDatum {
-    // ----------------------------------------------------------------------
-    //
-    // manipulating the SNP information
-    //
-    // ----------------------------------------------------------------------
-    /** Location of this variant on the reference (on the forward strand).
-     *
-     * @return
-     */
-    GenomeLoc getLocation();
-
-    /** Returns bases in the reference allele as a String. String can be empty (as in insertion into
-     * the reference), can contain a single character (as in SNP or one-base deletion), or multiple characters
-     * (for longer indels).
-     *
-     * @return reference allele, forward strand
-     */
-    String getRefBasesFWD();
-
-    /** Returns reference (major) allele base for a SNP variant as a character; should throw IllegalStateException
-     * if variant is not a SNP.
-     *
-     * @return reference base on the forward strand
-     */
-    char getRefSnpFWD() throws IllegalStateException;
-
-    /** Returns bases in the alternative allele as a String. String can be empty (as in deletion from
-     * the reference), can contain a single character (as in SNP or one-base insertion), or multiple characters
-     * (for longer indels).
-     *
-     * @return alternative allele, forward strand
-     */
-    String getAltBasesFWD();
-
-    /** Returns alternative (minor) allele base for a SNP variant as a character; should throw IllegalStateException
-     * if variant is not a SNP.
-     *
-     * @return alternative allele base on the forward starnd
-     */
-    char getAltSnpFWD() throws IllegalStateException;
-    
-    /** Returns true if all observed alleles are reference alleles. All is<Variant> methods (where Variant=SNP,Insertion, etc) should
-     * return false at such site to ensure consistency. This method is included for use with genotyping calls (isGenotype()==true), it makes
-     * no sense for, e.g. dbSNP and should return false for the latter.
-     * @return
-     */
-    boolean isReference();
-
-    /** Is this variant a SNP?
-     *
-     * @return true or false
-     */
-    boolean isSNP();
-
-    /** Is this variant an insertion? The contract requires isIndel() to return true
-     * if this method returns true.
-     *
-     * @return true or false
-     */
-    boolean isInsertion();
-
-    /** Is this variant a deletion? The contract requires isIndel() to return true
-     * if isDeletion() returns true.
-     *
-     * @return true or false
-     */
-    boolean isDeletion();
-
-    /** Is this variant an insertion or a deletion? The contract requires
-     * this to be true if either isInsertion() or isDeletion() returns true. However,
-     * this method is currently allowed to return true even if neither of isInsertion()
-     * and isDeletion() does.
-     * @return
-     */
-    boolean isIndel();
-
-    /** Returns minor allele frequency.
-     *
-     * @return
-     */
-    double getMAF() ;
-
-    /** Returns heterozygosity, a more accessible general feature of a variant.
-     *
-     * @return
-     */
-    double getHeterozygosity() ;
-    
-    /** Is this variant an actual genotype (such as individual call from sequencing, HapMap chip etc), or
-     * population allelic variant (call from pooled sequencing, dbSNP site etc). Only if variant is a genotype, there
-     * is a meaningful question of, e.g., whether it is a het, or homogeneous non-ref.
-     *
-      * @return true if this variant is an actual genotype.
-     */
-    boolean isGenotype();
-
-    /** Returns phred-mapped confidence in variation event (e.g. MAQ's SNP confidence, or AlleleCaller's best vs. ref).
-     *
-     * @return
-     */
-    double getVariationConfidence();
-
-    /** Returns phred-mapped confidence in called genotype (e.g. MAQ's consensus confidence, or AlleleCaller's
-     * best vs next-best.
-     * @return
-     */
-    double getConsensusConfidence();
-
-    /** Returns actual observed genotype. Allowed to return more than two alleles (@see #getPloidy()). If this variant
-     * is not a genotype, should throw an IllegalStateException.
-     * @return
-     */
-    List<String> getGenotype() throws IllegalStateException;
-
-    /** Return actual number of observed alleles (chromosomes) in the genotype. If this variant is not a genotype,
-     * should throw IllegalStateException.
-     * @return
-     */
-    int getPloidy() throws IllegalStateException;
-    
-    /** Returns true if the site has at most two known or observed alleles (ref and non-ref), or false if there are > 2 allelic variants known or observed. When
-     * the implementing class is a genotype, alleles should be always counted including the reference allele whether it was observed in the particular
-     * individual or not: i.e. if the reference is 'C', then both 'CA' and 'AA' genotypes must be reported as bi-allelic, while 'AT' is <i>not</i> bi-allelic (since there are
-     * two allelic variants, 'A' and 'T' <i>in addition</i> to the (known) reference variant 'C'). 
-     * @return
-     */
-    boolean isBiallelic();
-
-    /** returns the length of the variant.  For SNPs this is just 1.
-     */
-    int length();
-
-}

java/src/org/broadinstitute/sting/gatk/refdata/RodGeliText.java

@@ -37,7 +37,7 @@ import java.util.ArrayList;
 import java.util.Arrays;
 import java.util.List;
 
-public class RodGeliText extends BasicReferenceOrderedDatum implements VariationRod, VariantBackedByGenotype, AllelicVariant {
+public class RodGeliText extends BasicReferenceOrderedDatum implements VariationRod, VariantBackedByGenotype {
     public enum Genotype_Strings {
         AA, AC, AG, AT, CC, CG, CT, GG, GT, TT
     }

java/src/org/broadinstitute/sting/gatk/refdata/RodGenotypeChipAsGFF.java

@@ -18,7 +18,7 @@ import java.util.regex.Pattern;
  * Time: 10:47:14 AM
  * To change this template use File | Settings | File Templates.
  */
-public class RodGenotypeChipAsGFF extends BasicReferenceOrderedDatum implements AllelicVariant, VariationRod, VariantBackedByGenotype {
+public class RodGenotypeChipAsGFF extends BasicReferenceOrderedDatum implements VariationRod, VariantBackedByGenotype {
     private String contig, source, feature, strand, frame;
     private long start, stop;
     private double score;

java/src/org/broadinstitute/sting/gatk/refdata/SNPCallFromGenotypes.java

@@ -1,5 +1,7 @@
 package org.broadinstitute.sting.gatk.refdata;
 
+import org.broadinstitute.sting.utils.genotype.Variation;
+
 import java.util.List;
 
 /**
@@ -8,7 +10,7 @@ import java.util.List;
  * chr1:1104841 C N 0.000000 -69.937928 -69.937928 0.000000 0.000000 324.000000 162 0 0
  * chr1:1104842 A N 0.000000 -84.816002 -84.816002 0.000000 0.000000 324.000000 162 0 0
  */
-public interface SNPCallFromGenotypes extends AllelicVariant {
+public interface SNPCallFromGenotypes extends Variation {
     public int nIndividuals();
 
     public int nHomRefGenotypes();

java/src/org/broadinstitute/sting/gatk/refdata/SangerSNPROD.java

@@ -1,10 +1,11 @@
 package org.broadinstitute.sting.gatk.refdata;
 
-import java.util.*;
-
 import org.broadinstitute.sting.utils.GenomeLoc;
 import org.broadinstitute.sting.utils.GenomeLocParser;
 
+import java.util.Arrays;
+import java.util.List;
+
 public class SangerSNPROD extends TabularROD implements SNPCallFromGenotypes {
     public SangerSNPROD(final String name) {
         super(name);
@@ -14,15 +15,71 @@ public class SangerSNPROD extends TabularROD implements SNPCallFromGenotypes {
         loc = GenomeLocParser.createGenomeLoc(this.get("0"), Long.parseLong(this.get("1")));
         return loc;
     }
+
+    /**
+     * get the reference base(s) at this position
+     *
+     * @return the reference base or bases, as a string
+     */
+    @Override
+    public String getReference() {
+         return String.valueOf(getRefBasesFWD().charAt(0));
+    }
+
     public String getRefBasesFWD() { return this.get("2"); }
     public char getRefSnpFWD() throws IllegalStateException { return getRefBasesFWD().charAt(0); }
     public String getAltBasesFWD() { return this.get("3"); }
     public char getAltSnpFWD() throws IllegalStateException { return getAltBasesFWD().charAt(0); }
     public boolean isReference()   { return getVariationConfidence() < 0.01; }
+
+    /**
+     * get the frequency of this variant, if we're a variant.  If we're reference this method
+     * should return 0.
+     *
+     * @return double with the stored frequency
+     */
+    @Override
+    public double getNonRefAlleleFrequency() {
+       return this.getMAF();
+    }
+
+    /**
+     * A convenience method, for switching over the variation type
+     *
+     * @return the VARIANT_TYPE of the current variant
+     */
+    @Override
+    public VARIANT_TYPE getType() {
+        if (isReference()) return VARIANT_TYPE.REFERENCE;
+        else return VARIANT_TYPE.SNP;
+    }
+
     public boolean isSNP()         { return ! isReference(); }
     public boolean isInsertion()   { return false; }
     public boolean isDeletion()    { return false; }
     public boolean isIndel()       { return false; }
+
+    /**
+     * gets the alternate base is the case of a SNP.  Throws an IllegalStateException if we're not a SNP
+     * of
+     *
+     * @return a char, representing the alternate base
+     */
+    @Override
+    public char getAlternativeBaseForSNP() {
+        return this.getAltSnpFWD();
+    }
+
+    /**
+     * gets the reference base is the case of a SNP.  Throws an IllegalStateException if we're not a SNP
+     *
+     * @return a char, representing the alternate base
+     */
+    @Override
+    public char getReferenceForSNP() {
+        return this.getRefSnpFWD();
+    }
+
     public double getMAF()         { return -1; }
     public double getHeterozygosity() { return -1; }
     public boolean isGenotype()    { return false; }
@@ -31,6 +88,37 @@ public class SangerSNPROD extends TabularROD implements SNPCallFromGenotypes {
     public List<String> getGenotype() throws IllegalStateException { throw new IllegalStateException(); }
     public int getPloidy() throws IllegalStateException { return 2; }
     public boolean isBiallelic()   { return true; }
+
+    /**
+     * get the -1 * (log 10 of the error value)
+     *
+     * @return the postive number space log based error estimate
+     */
+    @Override
+    public double getNegLog10PError() {
+        return this.getVariationConfidence();
+    }
+
+    /**
+     * gets the alternate base.  Use this method if we're biallelic
+     *
+     * @return
+     */
+    @Override
+    public String getAlternateBases() {
+        return this.get("3");
+    }
+
+    /**
+     * gets the alternate bases.  Use this method if the allele count is greater then 2 (not biallelic)
+     *
+     * @return
+     */
+    @Override
+    public List<String> getAlternateBaseList() {
+        return Arrays.asList(this.get("3"));
+    }
+
     public int length() { return 1; }
 
     // SNPCallFromGenotypes interface

java/src/org/broadinstitute/sting/gatk/refdata/SequenomROD.java

@@ -2,10 +2,13 @@ package org.broadinstitute.sting.gatk.refdata;
 
 import org.broadinstitute.sting.utils.GenomeLoc;
 import org.broadinstitute.sting.utils.GenomeLocParser;
+import org.broadinstitute.sting.utils.StingException;
+import org.broadinstitute.sting.utils.genotype.Variation;
 
-import java.util.*;
+import java.util.Arrays;
+import java.util.List;
 
-public class SequenomROD extends TabularROD implements AllelicVariant {
+public class SequenomROD extends TabularROD implements Variation {
 
     public SequenomROD(String name) {
         super(name);
@@ -16,29 +19,102 @@ public class SequenomROD extends TabularROD implements AllelicVariant {
         return GenomeLocParser.createGenomeLoc(this.get("0"), pos);
     }
 
+    /**
+     * get the reference base(s) at this position
+     *
+     * @return the reference base or bases, as a string
+     */
+    @Override
+    public String getReference() {
+        return "";
+    }
+
     public List<String> getFWDAlleles() {
         return Arrays.asList(this.get("2"));
     }
 
-    public String getFWDRefBases() { return ""; }
     public String getAltBasesFWD() { return getFWDAlleles().get(0); }
-    public String getRefBasesFWD() { return ""; }
-    public char getRefSnpFWD() { return 'N'; }
-    public char getAltSnpFWD() { return getAltBasesFWD().charAt(0); }
-    public char getRef() { return 'N'; }
-    public List<String> getGenotype() { return getFWDAlleles(); }
-    public boolean isGenotype() { return false; }
-    public boolean isPointGenotype() { return true; }
-    public boolean isIndelGenotype() { return false; }
+
+    /**
+     * get the frequency of this variant, if we're a variant.  If we're reference this method
+     * should return 0.
+     *
+     * @return double with the stored frequency
+     */
+    @Override
+    public double getNonRefAlleleFrequency() {
+        return 0.0;
+    }
+
+    /**
+     * A convenience method, for switching over the variation type
+     *
+     * @return the VARIANT_TYPE of the current variant
+     */
+    @Override
+    public VARIANT_TYPE getType() {
+        return VARIANT_TYPE.SNP;
+    }
+
     public boolean isSNP() { return true; }
     public boolean isReference() { return false; }
     public boolean isInsertion() { return false; }
     public boolean isDeletion() { return false; }
     public boolean isIndel() { return false; }
-    public double getVariantConfidence() { return 0.0; }
-    public double getVariationConfidence() { return 0.0; }
-    public double getConsensusConfidence() { return 0.0; }
+
+    /**
+     * gets the alternate base is the case of a SNP.  Throws an IllegalStateException if we're not a SNP
+     * of
+     *
+     * @return a char, representing the alternate base
+     */
+    @Override
+    public char getAlternativeBaseForSNP() {
+        return 'N';
+    }
+
+    /**
+     * gets the reference base is the case of a SNP.  Throws an IllegalStateException if we're not a SNP
+     *
+     * @return a char, representing the alternate base
+     */
+    @Override
+    public char getReferenceForSNP() {
+        return getAltBasesFWD().charAt(0);
+    }
+
     public boolean isBiallelic() { return true; }
+
+    /**
+     * get the -1 * (log 10 of the error value)
+     *
+     * @return the postive number space log based error estimate
+     */
+    @Override
+    public double getNegLog10PError() {
+        return 0.0;
+    }
+
+    /**
+     * gets the alternate bases.  Use this method if we're biallelic
+     *
+     * @return
+     */
+    @Override
+    public String getAlternateBases() {
+        return getAltBasesFWD();
+    }
+
+    /**
+     * gets the alternate bases.  Use this method if the allele count is greater then 2 (not biallelic)
+     *
+     * @return
+     */
+    @Override
+    public List<String> getAlternateBaseList() {
+        throw new StingException("SequenomRod is not biallelic");
+    }
+
     public boolean isHom() { return false; }
     public boolean isHet() { return false; }
     public double getHeterozygosity() { return 0.0; }

java/src/org/broadinstitute/sting/gatk/refdata/SimpleIndelROD.java

@@ -6,7 +6,7 @@ import org.broadinstitute.sting.utils.GenomeLocParser;
 import java.util.Arrays;
 import java.util.List;
 
-public class SimpleIndelROD extends TabularROD implements Genotype, AllelicVariant, VariationRod {
+public class SimpleIndelROD extends TabularROD implements Genotype, VariationRod {
 
     private boolean KGENOMES_FORMAT = false, checkedFormat = false;
 

java/src/org/broadinstitute/sting/gatk/refdata/rodDbSNP.java

@@ -20,7 +20,7 @@ import java.util.List;
  * Time: 10:47:14 AM
  * To change this template use File | Settings | File Templates.
  */
-public class rodDbSNP extends BasicReferenceOrderedDatum implements VariationRod, VariantBackedByGenotype, AllelicVariant {
+public class rodDbSNP extends BasicReferenceOrderedDatum implements VariationRod, VariantBackedByGenotype {
     public GenomeLoc loc;       // genome location of SNP
                                 // Reference sequence chromosome or scaffold
                                 // Start and stop positions in chrom

java/src/org/broadinstitute/sting/gatk/walkers/indels/SNPClusterWalker.java

@@ -2,8 +2,8 @@ package org.broadinstitute.sting.gatk.walkers.indels;
 
 import org.broadinstitute.sting.gatk.contexts.AlignmentContext;
 import org.broadinstitute.sting.gatk.contexts.ReferenceContext;
-import org.broadinstitute.sting.gatk.refdata.AllelicVariant;
 import org.broadinstitute.sting.gatk.refdata.RefMetaDataTracker;
+import org.broadinstitute.sting.gatk.refdata.VariationRod;
 import org.broadinstitute.sting.gatk.walkers.*;
 import org.broadinstitute.sting.utils.GenomeLoc;
 import org.broadinstitute.sting.utils.GenomeLocParser;
@@ -11,7 +11,7 @@ import org.broadinstitute.sting.utils.cmdLine.Argument;
 import org.broadinstitute.sting.utils.genotype.Variation;
 
 @WalkerName("SNPClusters")
-@Requires(value={DataSource.REFERENCE},referenceMetaData={@RMD(name="snps",type=AllelicVariant.class)})
+@Requires(value={DataSource.REFERENCE},referenceMetaData={@RMD(name="snps",type= VariationRod.class)})
 public class SNPClusterWalker extends RefWalker<GenomeLoc, GenomeLoc> {
     @Argument(fullName="windowSize", shortName="window", doc="window size for calculating clusters", required=false)
     int windowSize = 10;

java/src/org/broadinstitute/sting/playground/gatk/walkers/MendelianInheritanceWalker.java

@@ -1,20 +1,20 @@
 package org.broadinstitute.sting.playground.gatk.walkers;
 
 
-import java.util.List;
-
 import org.apache.log4j.Logger;
 import org.broadinstitute.sting.gatk.contexts.AlignmentContext;
 import org.broadinstitute.sting.gatk.contexts.ReferenceContext;
+import org.broadinstitute.sting.gatk.refdata.Genotype;
 import org.broadinstitute.sting.gatk.refdata.RefMetaDataTracker;
 import org.broadinstitute.sting.gatk.refdata.ReferenceOrderedDatum;
-import org.broadinstitute.sting.gatk.refdata.Genotype;
 import org.broadinstitute.sting.gatk.walkers.RefWalker;
 import org.broadinstitute.sting.playground.utils.GenotypingCallStats;
 import org.broadinstitute.sting.playground.utils.TrioConcordanceRecord;
 import org.broadinstitute.sting.utils.GenotypeUtils;
 import org.broadinstitute.sting.utils.cmdLine.Argument;
 
+import java.util.List;
+
 //@Requires(value=DataSource.REFERENCE,referenceMetaData={@RMD(name="mother",type=rodSAMPileup.class),
 //                                                        @RMD(name="father",type=rodSAMPileup.class),
 //                                                        @RMD(name="daughter",type=rodSAMPileup.class)})

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

@@ -83,7 +83,7 @@ public class HardyWeinbergEquilibrium extends BasicVariantAnalysis implements Po
 
         if ( false ) {
             System.out.printf("HWE-violation at %s %f < %f %1.2f %5d %5d %5d %5d %5d %.2e %.2e %.2e => %.6e [%s]%n",
-                    call.getLocation(), p, threshold, call.getMAF(), nAA, nAa, naa, nA, n, m, b, tosses, p, call);
+                    call.getLocation(), p, threshold, call.getNonRefAlleleFrequency(), nAA, nAa, naa, nA, n, m, b, tosses, p, call);
             System.out.printf("(factorial(%d) / (factorial(%d) * factorial(%d) * factorial(%d))) / choose(%d, %d) * 2^%d - %f < 1e-3%n",
                     nAA + nAa + naa, nAA, nAa, naa, 2 * n, nA, nAa, p);
         }

java/src/org/broadinstitute/sting/utils/GenotypeUtils.java

@@ -1,12 +1,12 @@
 package org.broadinstitute.sting.utils;
 
-import java.util.Arrays;
-import java.util.List;
-
 import org.broadinstitute.sting.gatk.refdata.Genotype;
 import org.broadinstitute.sting.gatk.refdata.GenotypeList;
 import org.broadinstitute.sting.gatk.refdata.ReferenceOrderedDatum;
-import org.broadinstitute.sting.gatk.refdata.AllelicVariant;
+import org.broadinstitute.sting.utils.genotype.Variation;
+
+import java.util.Arrays;
+import java.util.List;
 
 /** Holds useful utility methods and auxiliary default classes for working with Genotype objects
  * 
@@ -77,15 +77,19 @@ public class GenotypeUtils {
 		else throw new StingException("track "+rod.getName()+" is not a Genotype or GenotypeList");
 	}
 
-    public static boolean isHet(AllelicVariant var) {
+    public static boolean isHet(Variation var) {
         if ( var instanceof Genotype )
             return ((Genotype)var).isHet();
 
-        List<String> genotype = var.getGenotype();
-        if ( genotype.size() < 1 )
+        String genotype = var.getAlternateBases();
+        if ( genotype.length() < 1 )
             return false;
 
-        return genotype.get(0).charAt(0) != genotype.get(0).charAt(1);
+        char first = genotype.charAt(0);
+        for (char base : genotype.toCharArray()) {
+            if (base != first) return true;
+        }
+        return false;
     }
 
 	/** This class represents a "default" indel-type genotype with homozygous reference (i.e. confidently no indel)

java/test/org/broadinstitute/sting/utils/GenotypeUtilsTest.java

@@ -0,0 +1,52 @@
+package org.broadinstitute.sting.utils;
+
+import org.broadinstitute.sting.BaseTest;
+import org.broadinstitute.sting.utils.fasta.IndexedFastaSequenceFile;
+import org.broadinstitute.sting.utils.genotype.BasicVariation;
+import org.junit.Assert;
+import org.junit.BeforeClass;
+import org.junit.Test;
+
+import java.io.File;
+import java.io.FileNotFoundException;
+
+
+/**
+ * 
+ * @author aaron 
+ * 
+ * Class GenotypeUtilsTest
+ *
+ * a test class for the various function in the genotype utils class
+ */
+public class GenotypeUtilsTest extends BaseTest {
+
+    private static IndexedFastaSequenceFile seq;
+    private static File vcfFile = new File("/humgen/gsa-scr1/GATK_Data/Validation_Data/vcfexample.vcf");
+
+    @BeforeClass
+    public static void beforeTests() {
+        try {
+            seq = new IndexedFastaSequenceFile(new File("/broad/1KG/reference/human_b36_both.fasta"));
+        } catch (FileNotFoundException e) {
+            throw new StingException("unable to load the sequence dictionary");
+        }
+        GenomeLocParser.setupRefContigOrdering(seq);
+
+    }
+
+    /**
+     * make sure that the variation is a het
+     */
+    @Test
+    public void isHetTest() {
+        GenomeLoc loc = GenomeLocParser.createGenomeLoc(0,1,2);
+        BasicVariation var = new BasicVariation("AA", "A", 0, loc,0.0);
+        Assert.assertTrue(!GenotypeUtils.isHet(var));
+        BasicVariation var2 = new BasicVariation("AG", "A", 0, loc,0.0);
+        Assert.assertTrue(GenotypeUtils.isHet(var2));
+        BasicVariation var3 = new BasicVariation("GG", "A", 0, loc,0.0);
+        Assert.assertTrue(!GenotypeUtils.isHet(var3));
+    }
+
+}