Merge pull request #300 from biowdl/fix_gridss_tbi

run tabix if vcf index is missing in gridss

CHANGELOG.md

@@ -9,7 +9,9 @@ that users understand how the changes affect the new version.
 -->
 version 5.1.0-dev
 ---------------------------
-+ Add a script to subtract UMI's from the read name and add them as 
++ The GRIDSS task will now run tabix separately if GRIDSS doesn't
+  produce a vcf index.
++ Add a script to subtract UMI's from the read name and add them as
   a BAM tag for each BAM record. The script is in umi.BamReadNameToUmiTag.
 + Add fgbio.AnnotateBamWithUmis.
 + Add picard.UmiAwareMarkDuplicatesWithMateCigar.
@@ -32,7 +34,7 @@ version 5.1.0-dev
   + Sage
   + VirusInterpreter
 + Added a task for VirusBreakend.
-+ Added a task for GridssAnnotateVcfRepeatmasker. 
++ Added a task for GridssAnnotateVcfRepeatmasker.
 + Bumped GRIDSS version to 2.12.2.
 + Adjusted GRIDSS runtime settings.
 + Added optional inputs to GRIDSS:
@@ -151,7 +153,7 @@ version 4.0.0
 + Picard MergeVcf now uses compression level 1 by default.
 + bwa mem, bwa mem+kit and hisat2 have their samtools sort threads tweaked. The
   number of threads is now related to the number of threads on the aligner.
-  Using more threads reduces the chance of the samtools sort pipe getting 
+  Using more threads reduces the chance of the samtools sort pipe getting
   blocked if it's full.
 + Renamed a few inputs in centrifuge.wdl, isoseq3.wdl, talon.wdl,
   transcriptclean.wdl to be more descriptive.

gridss.wdl

@@ -116,6 +116,12 @@ task GRIDSS {
         ~{normalBam} \
         ~{tumorBam}
         samtools index ~{outputPrefix}_assembly.bam ~{outputPrefix}_assembly.bai
+
+        # For some reason the VCF index is sometimes missing
+        if [ ! -e ~{outputPrefix}.vcf.gz.tbi ]
+          then
+            tabix ~{outputPrefix}.vcf.gz
+        fi
     }
 
     output {