Support fasta.gz in GATKSparkTool (broadinstitute#5290)

* fasta.gz references can now be used in GATKSparkTools * fixes broadinstitute#5258
EdwardDixon · Nov 9, 2018 · da1bb69 · da1bb69
1 parent eccb5b1
commit da1bb69
Show file tree

Hide file tree

Showing 6 changed files with 18 additions and 0 deletions.
diff --git a/src/main/java/org/broadinstitute/hellbender/engine/spark/GATKSparkTool.java b/src/main/java/org/broadinstitute/hellbender/engine/spark/GATKSparkTool.java
@@ -3,6 +3,7 @@
 import htsjdk.samtools.SAMFileHeader;
 import htsjdk.samtools.SAMSequenceDictionary;
 import htsjdk.samtools.reference.ReferenceSequenceFileFactory;
+import htsjdk.samtools.util.GZIIndex;
 import htsjdk.samtools.util.IOUtil;
 import htsjdk.variant.vcf.VCFHeaderLine;
 import org.apache.spark.api.java.JavaRDD;
@@ -588,6 +589,7 @@ protected static String addReferenceFilesForSpark(JavaSparkContext ctx, String r
         Path referencePath = IOUtils.getPath(referenceFile);
         Path indexPath = ReferenceSequenceFileFactory.getFastaIndexFileName(referencePath);
         Path dictPath = ReferenceSequenceFileFactory.getDefaultDictionaryForReferenceSequence(referencePath);
+        Path gziPath = GZIIndex.resolveIndexNameForBgzipFile(referencePath);
 
         ctx.addFile(referenceFile);
         if (Files.exists(indexPath)) {
@@ -596,6 +598,9 @@ protected static String addReferenceFilesForSpark(JavaSparkContext ctx, String r
         if (Files.exists(dictPath)) {
             ctx.addFile(dictPath.toUri().toString());
         }
+        if (Files.exists(gziPath)) {
+            ctx.addFile(gziPath.toUri().toString());
+        }
 
         return referencePath.getFileName().toString();
     }

diff --git a/src/test/java/org/broadinstitute/hellbender/GATKBaseTest.java b/src/test/java/org/broadinstitute/hellbender/GATKBaseTest.java
@@ -49,6 +49,8 @@ public abstract class GATKBaseTest extends BaseTest {
     // 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";
 
+    public static final String b37_reference_20_21_gz = largeFileTestDir + "human_g1k_v37.20.21.fasta.gz";
+
     public static final String b37_2bit_reference_20_21 = largeFileTestDir + "human_g1k_v37.20.21.2bit";
 
     public static final String b37_reference_20_21_img = largeFileTestDir + "human_g1k_v37.20.21.fasta.img";

diff --git a/.../java/org/broadinstitute/hellbender/tools/spark/BaseRecalibratorSparkIntegrationTest.java b/.../java/org/broadinstitute/hellbender/tools/spark/BaseRecalibratorSparkIntegrationTest.java
@@ -63,6 +63,7 @@ public Object[][] createBQSRTestData() {
         final String localResources =  getResourceDir();
 
         final String GRCh37Ref_chr2021 = b37_reference_20_21;
+        final String GRCh37Ref_chr2021_gz = b37_reference_20_21_gz;
         final String hiSeqBam_chr20 = localResources + WGS_B37_CH20_1M_1M1K_BAM;
         final String hiSeqBam_1read = localResources + "overlappingRead.bam";
         final String dbSNPb37_chr20 = localResources + DBSNP_138_B37_CH20_1M_1M1K_VCF;
@@ -87,6 +88,7 @@ public Object[][] createBQSRTestData() {
                 // local input/computation/reference
                 {new BQSRTest(GRCh37Ref_chr2021, hiSeqBam_1read, dbSNPb37_chr2021, "-indels --enable-baq", getResourceDir() + BQSRTestData.EXPECTED_WGS_B37_CH20_1READ_RECAL)},
                 {new BQSRTest(GRCh37Ref_chr2021, hiSeqBam_chr20, dbSNPb37_chr20, "-indels --enable-baq", getResourceDir() + BQSRTestData.EXPECTED_WGS_B37_CH20_1M_1M1K_RECAL)},
+                {new BQSRTest(GRCh37Ref_chr2021_gz, hiSeqBam_chr20, dbSNPb37_chr20, "-indels --enable-baq", getResourceDir() + BQSRTestData.EXPECTED_WGS_B37_CH20_1M_1M1K_RECAL)},
                 {new BQSRTest(GRCh37Ref_chr2021, hiSeqBam_chr20, dbSNPb37_chr20, "", getResourceDir() + BQSRTestData.EXPECTED_WGS_B37_CH20_1M_1M1K_NOINDEL_NOBAQ_RECAL)},
                 {new BQSRTest(GRCh37Ref_chr2021, hiSeqBam_chr20, dbSNPb37_chr20, "-indels --enable-baq --indels-context-size 4", getResourceDir() + BQSRTestData.EXPECTED_WGS_B37_CH20_1M_1M1K_INDELS_CONTEXT_SIZE_4_RECAL)},
                 {new BQSRTest(GRCh37Ref_chr2021, hiSeqBam_chr20, dbSNPb37_chr20, "-indels --enable-baq --low-quality-tail 5", getResourceDir() + BQSRTestData.EXPECTED_WGS_B37_CH20_1M_1M1K_LOW_QUALITY_TAIL_5_RECAL)},

diff --git a/src/test/resources/large/human_g1k_v37.20.21.fasta.gz b/src/test/resources/large/human_g1k_v37.20.21.fasta.gz
diff --git a/src/test/resources/large/human_g1k_v37.20.21.fasta.gz.fai b/src/test/resources/large/human_g1k_v37.20.21.fasta.gz.fai
diff --git a/src/test/resources/large/human_g1k_v37.20.21.fasta.gz.gzi b/src/test/resources/large/human_g1k_v37.20.21.fasta.gz.gzi