From 2b471cabe37b986049780ed4bbb4c9838be9a321 Mon Sep 17 00:00:00 2001
From: David Roazen <droazen@broadinstitute.org>
Date: Mon, 20 Jul 2015 12:42:59 -0400
Subject: [PATCH] Fix BQSR Dataflow test that was failing due to lack of
 sequence dictionary validation

Now that sequence dictionary validation is in, we can re-enable this test,
which was previously failing with a java.lang.OutOfMemoryError due to lack
of upfront validation of the reads vs. reference sequence dictionaries.
---
 .../walkers/bqsr/BaseRecalibratorDataflow.java   | 16 ++++++++++++----
 .../BaseRecalibratorDataflowIntegrationTest.java |  5 ++---
 2 files changed, 14 insertions(+), 7 deletions(-)

diff --git a/src/main/java/org/broadinstitute/hellbender/dev/tools/walkers/bqsr/BaseRecalibratorDataflow.java b/src/main/java/org/broadinstitute/hellbender/dev/tools/walkers/bqsr/BaseRecalibratorDataflow.java
index 96626b1ec37..a1977e39c10 100644
--- a/src/main/java/org/broadinstitute/hellbender/dev/tools/walkers/bqsr/BaseRecalibratorDataflow.java
+++ b/src/main/java/org/broadinstitute/hellbender/dev/tools/walkers/bqsr/BaseRecalibratorDataflow.java
@@ -22,10 +22,7 @@
 import org.broadinstitute.hellbender.dev.BunnyLog;
 import org.broadinstitute.hellbender.dev.pipelines.bqsr.BaseRecalibratorDataflowUtils;
 import org.broadinstitute.hellbender.dev.pipelines.bqsr.DataflowReadFilter;
-import org.broadinstitute.hellbender.engine.FeatureDataSource;
-import org.broadinstitute.hellbender.engine.FeatureInput;
-import org.broadinstitute.hellbender.engine.FeatureManager;
-import org.broadinstitute.hellbender.engine.ReadsDataSource;
+import org.broadinstitute.hellbender.engine.*;
 import org.broadinstitute.hellbender.engine.dataflow.DataflowCommandLineProgram;
 import org.broadinstitute.hellbender.engine.dataflow.ReadsSource;
 import org.broadinstitute.hellbender.engine.dataflow.coders.GATKReadCoder;
@@ -34,6 +31,7 @@
 import org.broadinstitute.hellbender.exceptions.UserException;
 import org.broadinstitute.hellbender.tools.recalibration.RecalibrationTables;
 import org.broadinstitute.hellbender.utils.IntervalUtils;
+import org.broadinstitute.hellbender.utils.SequenceDictionaryUtils;
 import org.broadinstitute.hellbender.utils.SimpleInterval;
 import org.broadinstitute.hellbender.utils.dataflow.BucketUtils;
 import org.broadinstitute.hellbender.utils.dataflow.DataflowUtils;
@@ -116,6 +114,7 @@ protected void setupPipeline(Pipeline pipeline) {
             BaseRecalibratorDataflowUtils.ensureReferenceIsReadable(pipeline.getOptions(), referencePath);
             baseRecalibratorWorker = BaseRecalibratorWorker.fromArgs(header, BRAC);
             baseRecalibratorWorker.checkClientArguments();
+            checkSequenceDictionaries();
 
             // 2. set up computation
             PCollection<RecalibrationTables> aggregated =
@@ -142,6 +141,15 @@ protected void setupPipeline(Pipeline pipeline) {
         }
     }
 
+    private void checkSequenceDictionaries() {
+        try ( final ReferenceDataSource refSource = new ReferenceDataSource(new File(referencePath)) ) {
+            final SAMSequenceDictionary refDictionary = refSource.getSequenceDictionary();
+            final SAMSequenceDictionary readsDictionary = header.getSequenceDictionary();
+
+            SequenceDictionaryUtils.validateDictionaries("reference", refDictionary, "reads", readsDictionary, true, null);
+        }
+    }
+
     @Override
     protected void afterPipeline(Pipeline p) {
         bunny.stepEnd("dataflow");
diff --git a/src/test/java/org/broadinstitute/hellbender/tools/walkers/bqsr/BaseRecalibratorDataflowIntegrationTest.java b/src/test/java/org/broadinstitute/hellbender/tools/walkers/bqsr/BaseRecalibratorDataflowIntegrationTest.java
index a309b2a8f92..6aa49055d81 100644
--- a/src/test/java/org/broadinstitute/hellbender/tools/walkers/bqsr/BaseRecalibratorDataflowIntegrationTest.java
+++ b/src/test/java/org/broadinstitute/hellbender/tools/walkers/bqsr/BaseRecalibratorDataflowIntegrationTest.java
@@ -238,8 +238,7 @@ public void testBQSRFailWithoutDBSNP() throws IOException {
         spec.executeTest("testBQSRFailWithoutDBSNP", this);
     }
 
-    // TODO: re-enable this once sequence dictionary validation is in
-    @Test(enabled = false)
+    @Test
     public void testBQSRFailWithIncompatibleReference() throws IOException {
         final String resourceDir =  getTestDataDir() + "/" + "BQSR" + "/";
 
@@ -250,7 +249,7 @@ public void testBQSRFailWithIncompatibleReference() throws IOException {
         IntegrationTestSpec spec = new IntegrationTestSpec(
                 params.getCommandLine(),
                 1,
-                UserException.MissingContigInSequenceDictionary.class);
+                UserException.IncompatibleSequenceDictionaries.class);
         spec.executeTest("testBQSRFailWithIncompatibleReference", this);
     }
 }