make sure we close the error log stream in CommandLineProgram if it's opened; unit tests and clean-up for BasicVariation

git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2241 348d0f76-0448-11de-a6fe-93d51630548a

java/src/org/broadinstitute/sting/utils/cmdLine/CommandLineProgram.java

@@ -261,6 +261,7 @@ public abstract class CommandLineProgram {
             throw new RuntimeException(e);
         }
         clp.generateErrorLog(stream, e);
+        stream.close();
     }
 
     /**

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

@@ -5,7 +5,9 @@ import org.broadinstitute.sting.utils.StingException;
 import org.broadinstitute.sting.utils.Utils;
 
 import java.util.ArrayList;
+import java.util.HashSet;
 import java.util.List;
+import java.util.Set;
 
 /**
  * User: aaron
@@ -65,14 +67,14 @@ public class BasicVariation implements Variation {
      */
     @Override
     public VARIANT_TYPE getType() {
-        if (mLength != 0) return VARIANT_TYPE.INSERTION;
+        if (mLength > 0) return VARIANT_TYPE.INSERTION;
+        if (mLength < 0) return VARIANT_TYPE.DELETION;
         return (isSNP()) ? VARIANT_TYPE.SNP : VARIANT_TYPE.REFERENCE;
     }
 
     @Override
     public boolean isSNP() {
-        if (mLength == 0) return true;
+        return ((mLength == 0) && (new HashSet(getAlternateAlleleList()).size() == 1));
-        return false;
     }
 
     @Override
@@ -95,10 +97,15 @@ public class BasicVariation implements Variation {
         return (mRef);
     }
 
-    /** are we bi-allelic? */
+    /**
+     * are we bi-allelic? In this case we always
+     * count the reference as an allele
+     */
     @Override
     public boolean isBiallelic() {
-        return (getAlternateAlleleList().size() == 1);
+        // put the alternate alleles into a set, there may be duplicates (i.e. hom var)
+        Set<String> alleles = new HashSet(getAlternateAlleleList());
+        return (alleles.size() == 1); // if the alt list contained one unqiue non-ref base, we're biallelic
     }
 
     @Override
@@ -133,20 +140,16 @@ public class BasicVariation implements Variation {
     @Override
     public List<String> getAlleleList() {
         List<String> list = new ArrayList<String>();
-        if (this.mBases.contains(mRef)) list.add(mRef);
         for (char c : this.mBases.toCharArray())
-            if (c != Utils.stringToChar(mRef))
+            list.add(String.valueOf(c));
-                list.add(String.valueOf(c));
         return list;
     }
 
     @Override
     public boolean isReference() {
         if (mLength != 0) return false;
-        int refIndex = 0;
+        for (String str : getAlleleList())
-        for (char c : mBases.toCharArray()) {
+            if (!str.equals(mRef)) return false;
-            if (mRef.charAt(refIndex) != c) return false;
-        }
         return true;
     }
 
@@ -169,8 +172,8 @@ public class BasicVariation implements Variation {
     @Override
     public char getAlternativeBaseForSNP() {
         if (!this.isSNP()) throw new IllegalStateException("we're not a SNP");
-        if (!this.isBiallelic() || this.getAlternateAlleleList().size() != 1) throw new IllegalStateException("we're not biallelic");
+        if (!this.isBiallelic()) throw new IllegalStateException("we're not biallelic");
-        return Utils.stringToChar(this.getAlternateAlleleList().get(0));
+        return Utils.stringToChar((new HashSet<String>(getAlternateAlleleList())).iterator().next());
     }
 
     /**

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

@@ -0,0 +1,117 @@
+package org.broadinstitute.sting.utils.genotype;
+
+import org.broadinstitute.sting.BaseTest;
+import org.broadinstitute.sting.utils.GenomeLocParser;
+import org.broadinstitute.sting.utils.StingException;
+import org.broadinstitute.sting.utils.fasta.IndexedFastaSequenceFile;
+import org.junit.Assert;
+import org.junit.BeforeClass;
+import org.junit.Test;
+
+import java.io.File;
+import java.io.FileNotFoundException;
+
+/**
+ * Created by IntelliJ IDEA.
+ * User: aaron
+ * Date: Dec 2, 2009
+ * Time: 1:05:58 AM
+ * <p/>
+ * some quick tests for the BasicVariation class
+ */
+public class BasicVariationTest extends BaseTest {
+    private static IndexedFastaSequenceFile seq;
+
+    @BeforeClass
+    public static void beforeTests() {
+        try {
+            seq = new IndexedFastaSequenceFile(new File(seqLocation + "/references/Homo_sapiens_assembly18/v0/Homo_sapiens_assembly18.fasta"));
+        } catch (FileNotFoundException e) {
+            throw new StingException("unable to load the sequence dictionary");
+        }
+        GenomeLocParser.setupRefContigOrdering(seq);
+
+    }
+
+    @Test
+    public void testIsBiallelic() {
+        BasicVariation var = new BasicVariation("CC", "C", 0, GenomeLocParser.createGenomeLoc(1, 1, 1), 1.22);
+        Assert.assertTrue(!var.isBiallelic());
+        BasicVariation var2 = new BasicVariation("CA", "C", 0, GenomeLocParser.createGenomeLoc(1, 1, 1), 1.22);
+        Assert.assertTrue(var2.isBiallelic());
+        BasicVariation var3 = new BasicVariation("CC", "A", 0, GenomeLocParser.createGenomeLoc(1, 1, 1), 1.22);
+        Assert.assertTrue(var3.isBiallelic());
+    }
+
+    @Test
+    public void testVariantType() {
+        // test reference
+        BasicVariation var = new BasicVariation("CC", "C", 0, GenomeLocParser.createGenomeLoc(1, 1, 1), 1.22);
+        Assert.assertTrue(var.getType() == Variation.VARIANT_TYPE.REFERENCE);
+
+        // test SNP's
+        BasicVariation var2 = new BasicVariation("CA", "C", 0, GenomeLocParser.createGenomeLoc(1, 1, 1), 1.22);
+        Assert.assertTrue(var2.getType() == Variation.VARIANT_TYPE.SNP);
+        BasicVariation var3 = new BasicVariation("AA", "C", 0, GenomeLocParser.createGenomeLoc(1, 1, 1), 1.22);
+        Assert.assertTrue(var3.getType() == Variation.VARIANT_TYPE.SNP);
+
+        // test deletions
+        BasicVariation var4 = new BasicVariation("", "C", -10, GenomeLocParser.createGenomeLoc(1, 1, 1), 1.22);
+        Assert.assertTrue(var4.getType() == Variation.VARIANT_TYPE.DELETION);
+
+        // test insertions
+        BasicVariation var5 = new BasicVariation("ACACACACACA", "C", "ACACACACACA".length(), GenomeLocParser.createGenomeLoc(1, 1, 1), 1.22);
+        Assert.assertTrue(var5.getType() == Variation.VARIANT_TYPE.INSERTION);
+
+    }
+
+    @Test(expected = IllegalStateException.class)
+    public void testGetAlternativeBaseForSNPNotASNP() {
+        // test reference
+        BasicVariation var = new BasicVariation("CC", "C", 0, GenomeLocParser.createGenomeLoc(1, 1, 1), 1.22);
+        var.getAlternativeBaseForSNP();
+    }
+
+    @Test(expected = IllegalStateException.class)
+    public void testGetAlternativeBaseForSNPFromIndel() {
+        // test reference
+        BasicVariation var = new BasicVariation("ACACACACACA", "C", "ACACACACACA".length(), GenomeLocParser.createGenomeLoc(1, 1, 1), 1.22);
+        var.getAlternativeBaseForSNP();
+    }
+
+    @Test(expected = IllegalStateException.class)
+    public void testGetAlternativeBaseForSNPFromDel() {
+        // test reference
+        BasicVariation var = new BasicVariation("", "C", -10, GenomeLocParser.createGenomeLoc(1, 1, 1), 1.22);
+        var.getAlternativeBaseForSNP();
+    }
+
+    @Test
+    public void testGetAlternativeBaseForSNP() {
+        // test SNP's
+        BasicVariation var = new BasicVariation("CA", "C", 0, GenomeLocParser.createGenomeLoc(1, 1, 1), 1.22);
+        Assert.assertEquals('A', var.getAlternativeBaseForSNP());
+        var = new BasicVariation("AC", "C", 0, GenomeLocParser.createGenomeLoc(1, 1, 1), 1.22);
+        Assert.assertEquals('A', var.getAlternativeBaseForSNP());
+        var = new BasicVariation("AA", "C", 0, GenomeLocParser.createGenomeLoc(1, 1, 1), 1.22);
+        Assert.assertEquals('A', var.getAlternativeBaseForSNP());
+    }
+
+    @Test
+    public void testGetAlleleList() {
+        BasicVariation var = new BasicVariation("CA", "C", 0, GenomeLocParser.createGenomeLoc(1, 1, 1), 1.22);
+        Assert.assertTrue(var.getAlleleList().size() == 2);
+        Assert.assertTrue(var.getAlleleList().contains("C"));
+        Assert.assertTrue(var.getAlleleList().contains("A"));
+
+        var = new BasicVariation("AC", "C", 0, GenomeLocParser.createGenomeLoc(1, 1, 1), 1.22);
+        Assert.assertTrue(var.getAlleleList().size() == 2);
+        Assert.assertTrue(var.getAlleleList().contains("C"));
+        Assert.assertTrue(var.getAlleleList().contains("A"));
+
+        var = new BasicVariation("AA", "C", 0, GenomeLocParser.createGenomeLoc(1, 1, 1), 1.22);
+        Assert.assertTrue(var.getAlleleList().size() == 2);
+        Assert.assertTrue(var.getAlleleList().get(0).equals("A"));
+        Assert.assertTrue(var.getAlleleList().get(1).equals("A"));
+    }
+}