Fixed poor implementation of isRefSource() and isRefSink() among others.

There was already a note in the code about how wrong the implementation was.
The bad code was causing a single-node graph to get cleaned up into nothing when pruning tails.
Delivers PT #61069820.

protected/java/src/org/broadinstitute/sting/gatk/walkers/haplotypecaller/graphs/BaseGraph.java

@@ -95,10 +95,13 @@ public class BaseGraph<V extends BaseVertex, E extends BaseEdge> extends Default
      */
     public boolean isReferenceNode( final V v ) {
         if( v == null ) { throw new IllegalArgumentException("Attempting to test a null vertex."); }
-        for( final BaseEdge e : edgesOf(v) ) {
+
-            if( e.isRef() ) { return true; }
+        for ( final BaseEdge e : edgesOf(v) ) {
+            if ( e.isRef() ) { return true; }
         }
-        return false;
+
+        // edge case: if the graph only has one node then it's a ref node, otherwise it's not
+        return (vertexSet().size() == 1);
     }
 
     /**
@@ -154,62 +157,46 @@ public class BaseGraph<V extends BaseVertex, E extends BaseEdge> extends Default
         return v.getAdditionalSequence(isSource(v));
     }
 
-    /**
-     * @param e the edge to test
-     * @return  true if this edge is a reference source edge
-     */
-    public boolean isRefSource( final E e ) {
-        if( e == null ) { throw new IllegalArgumentException("Attempting to test a null edge."); }
-        for( final E edgeToTest : incomingEdgesOf(getEdgeSource(e)) ) {
-            if( edgeToTest.isRef() ) { return false; }
-        }
-        return true;
-    }
-
     /**
      * @param v the vertex to test
      * @return  true if this vertex is a reference source
      */
     public boolean isRefSource( final V v ) {
         if( v == null ) { throw new IllegalArgumentException("Attempting to test a null vertex."); }
-        for( final E edgeToTest : incomingEdgesOf(v) ) {
+
-            if( edgeToTest.isRef() ) { return false; }
+        // confirm that no incoming edges are reference edges
+        for ( final E edgeToTest : incomingEdgesOf(v) ) {
+            if ( edgeToTest.isRef() ) { return false; }
         }
-        return true;
+
+        // confirm that there is an outgoing reference edge
+        for ( final E edgeToTest : outgoingEdgesOf(v) ) {
+            if ( edgeToTest.isRef() ) { return true; }
+        }
+
+        // edge case: if the graph only has one node then it's a ref sink, otherwise it's not
+        return (vertexSet().size() == 1);
     }
 
     /**
-     * @param e the edge to test
-     * @return  true if this edge is a reference sink edge
-     */
-    public boolean isRefSink( final E e ) {
-        if( e == null ) { throw new IllegalArgumentException("Attempting to test a null edge."); }
-        for( final E edgeToTest : outgoingEdgesOf(getEdgeTarget(e)) ) {
-            if( edgeToTest.isRef() ) { return false; }
-        }
-        return true;
-    }
-
-    /**
-     * // TODO -- the logic of this test is just wrong
      * @param v the vertex to test
      * @return  true if this vertex is a reference sink
      */
     public boolean isRefSink( final V v ) {
         if( v == null ) { throw new IllegalArgumentException("Attempting to test a null vertex."); }
-        for( final E edgeToTest : outgoingEdgesOf(v) ) {
-            if( edgeToTest.isRef() ) { return false; }
-        }
-        return true;
-    }
 
-    /**
+        // confirm that no outgoing edges are reference edges
-     * Is this both a refsink node and a reference node
+        for ( final E edgeToTest : outgoingEdgesOf(v) ) {
-     * @param v a non-null vertex
+            if ( edgeToTest.isRef() ) { return false; }
-     * @return true if v is both a sink and a reference node
+        }
-     */
+
-    public boolean isRefNodeAndRefSink(final V v) {
+        // confirm that there is an incoming reference edge
-        return isRefSink(v) && isReferenceNode(v);
+        for ( final E edgeToTest : incomingEdgesOf(v) ) {
+            if ( edgeToTest.isRef() ) { return true; }
+        }
+
+        // edge case: if the graph only has one node then it's a ref source, otherwise it's not
+        return (vertexSet().size() == 1);
     }
 
     /**
@@ -217,7 +204,7 @@ public class BaseGraph<V extends BaseVertex, E extends BaseEdge> extends Default
      */
     public V getReferenceSourceVertex( ) {
         for( final V v : vertexSet() ) {
-            if( isReferenceNode(v) && isRefSource(v) ) {
+            if( isRefSource(v) ) {
                 return v;
             }
         }
@@ -229,7 +216,7 @@ public class BaseGraph<V extends BaseVertex, E extends BaseEdge> extends Default
      */
     public V getReferenceSinkVertex( ) {
         for( final V v : vertexSet() ) {
-            if( isReferenceNode(v) && isRefSink(v) ) {
+            if( isRefSink(v) ) {
                 return v;
             }
         }
@@ -490,7 +477,7 @@ public class BaseGraph<V extends BaseVertex, E extends BaseEdge> extends Default
         // Run through the graph and clean up singular orphaned nodes
         final List<V> verticesToRemove = new LinkedList<>();
         for( final V v : vertexSet() ) {
-            if( inDegreeOf(v) == 0 && outDegreeOf(v) == 0 ) {
+            if( inDegreeOf(v) == 0 && outDegreeOf(v) == 0 && !isRefSource(v) ) {
                 verticesToRemove.add(v);
             }
         }

protected/java/src/org/broadinstitute/sting/gatk/walkers/haplotypecaller/readthreading/ReadThreadingGraph.java

@@ -487,7 +487,7 @@ public class ReadThreadingGraph extends BaseGraph<MultiDeBruijnVertex, MultiSamp
         int attempted = 0;
         int nRecovered = 0;
         for ( final MultiDeBruijnVertex v : vertexSet() ) {
-            if ( outDegreeOf(v) == 0 && ! isRefNodeAndRefSink(v) ) {
+            if ( outDegreeOf(v) == 0 && ! isRefSink(v) ) {
                 attempted++;
                 nRecovered += recoverDanglingChain(v, pruneFactor);
             }

protected/java/test/org/broadinstitute/sting/gatk/walkers/haplotypecaller/HaplotypeCallerComplexAndSymbolicVariantsIntegrationTest.java

@@ -94,6 +94,6 @@ public class HaplotypeCallerComplexAndSymbolicVariantsIntegrationTest extends Wa
     @Test
     public void testHaplotypeCallerMultiSampleGGAMultiAllelic() {
         HCTestComplexGGA(NA12878_CHR20_BAM, "-L 20:133041-133161 -L 20:300207-300337",
-                "8e6a2002c59eafb78bdbf1db9660164b");
+                "f74d68cbc1ecb66a7128258e111cd030");
     }
 }

protected/java/test/org/broadinstitute/sting/gatk/walkers/haplotypecaller/graphs/BaseGraphUnitTest.java

@@ -95,7 +95,7 @@ public class BaseGraphUnitTest extends BaseTest {
     }
 
     @Test
-         public void testRemoveSingletonOrphanVertices() throws Exception {
+    public void testRemoveSingletonOrphanVertices() throws Exception {
         // all vertices in graph are connected
         final List<SeqVertex> kept = new LinkedList<SeqVertex>(graph.vertexSet());
         final SeqVertex rm1 = new SeqVertex("CAGT");
@@ -116,6 +116,52 @@ public class BaseGraphUnitTest extends BaseTest {
         Assert.assertFalse(graph.containsVertex(rm2));
     }
 
+    @Test
+    public void testRemoveSingletonOrphanVerticesOnSingleRefNode() throws Exception {
+        final SeqGraph original = new SeqGraph();
+        original.addVertex(v1);
+        original.removeSingletonOrphanVertices();
+        Assert.assertTrue(original.containsVertex(v1));
+        Assert.assertEquals(original.vertexSet().size(), 1);
+    }
+
+    @Test
+    public void testIsRefSourceAndSink() throws Exception {
+
+        final SeqGraph g = new SeqGraph();
+        g.addVertex(v1);
+        Assert.assertTrue(g.isRefSource(v1));
+        Assert.assertTrue(g.isRefSink(v1));
+        Assert.assertTrue(g.isReferenceNode(v1));
+
+        g.addVertices(v2, v3, v4, v5);
+        g.addEdge(v1, v2);
+        g.addEdge(v2, v3);
+        final BaseEdge refEdge = new BaseEdge(true, 1);
+        g.addEdge(v3, v4, refEdge);
+        g.addEdge(v4, v5);
+
+        Assert.assertFalse(g.isRefSource(v1));
+        Assert.assertFalse(g.isRefSink(v1));
+        Assert.assertFalse(g.isReferenceNode(v1));
+
+        Assert.assertFalse(g.isRefSource(v2));
+        Assert.assertFalse(g.isRefSink(v2));
+        Assert.assertFalse(g.isReferenceNode(v2));
+
+        Assert.assertTrue(g.isRefSource(v3));
+        Assert.assertFalse(g.isRefSink(v3));
+        Assert.assertTrue(g.isReferenceNode(v3));
+
+        Assert.assertFalse(g.isRefSource(v4));
+        Assert.assertTrue(g.isRefSink(v4));
+        Assert.assertTrue(g.isReferenceNode(v4));
+
+        Assert.assertFalse(g.isRefSource(v5));
+        Assert.assertFalse(g.isRefSink(v5));
+        Assert.assertFalse(g.isReferenceNode(v5));
+    }
+
     @Test
     public void testRemovePathsNotConnectedToRef() throws Exception {
         final SeqGraph graph = new SeqGraph();