Add full-sized B37 and HG38 references to our large test data (broadi…

…nstitute#5309)

Resolves broadinstitute#5111

src/test/java/org/broadinstitute/hellbender/GATKBaseTest.java

@@ -46,6 +46,14 @@ public abstract class GATKBaseTest extends BaseTest {
      */
     public static final String largeFileTestDir = new File(publicTestDir, "large").getAbsolutePath() + "/";
 
+    // The complete B37 human reference, including the Epstein-Barr contig, in fasta.gz format.
+    // Source: /seq/references/Homo_sapiens_assembly19/v1/ in the Broad Institute filesystem.
+    public static final String b37Reference = largeFileTestDir + "Homo_sapiens_assembly19.fasta.gz";
+
+    // The complete HG38 human reference, in fasta.gz format.
+    // Source: /seq/references/Homo_sapiens_assembly38/v0/ in the Broad Institute filesystem.
+    public static final String hg38Reference = largeFileTestDir + "Homo_sapiens_assembly38.fasta.gz";
+
     // All of chromosomes 20 and 21 from the b37 reference
     public static final String b37_reference_20_21 = largeFileTestDir + "human_g1k_v37.20.21.fasta";
 

src/test/java/org/broadinstitute/hellbender/engine/ReferenceDataSourceUnitTest.java

@@ -103,4 +103,28 @@ public void testQueryAndIterate(final Path testReference, final SimpleInterval i
             }
         }
     }
+
+    /**
+     * Test that we can successfully load and query our full-sized B37 reference.
+     */
+    @Test
+    public void testLoadAndQueryB37Reference() {
+        try (final ReferenceDataSource ref = new ReferenceFileSource(IOUtils.getPath(b37Reference))) {
+            Assert.assertEquals(ref.getSequenceDictionary().getSequences().size(), 85, "Wrong number of contigs in reference sequence dictionary");
+
+            Assert.assertTrue(Arrays.equals(ref.queryAndPrefetch("1", 10000000, 10000005).getBases(), new byte[]{ 'A', 'A', 'C', 'C', 'C', 'C' }), "Wrong reference bases returned for query on 1:10000000-10000005");
+        }
+    }
+
+    /**
+     * Test that we can successfully load and query our full-sized HG38 reference.
+     */
+    @Test
+    public void testLoadAndQueryHG38Reference() {
+        try (final ReferenceDataSource ref = new ReferenceFileSource(IOUtils.getPath(hg38Reference))) {
+            Assert.assertEquals(ref.getSequenceDictionary().getSequences().size(), 3366, "Wrong number of contigs in reference sequence dictionary");
+
+            Assert.assertTrue(Arrays.equals(ref.queryAndPrefetch("chr1", 10000000, 10000005).getBases(), new byte[]{ 'C', 'A', 'G', 'G', 'T', 'G' }), "Wrong reference bases returned for query on chr1:10000000-10000005");
+        }
+    }
 }