1 year-old bug in libgkl that has not been fixed affecting FilterAlignmentArtifacts and HaplotypeCaller

Follow

GE

• February 18, 2021 23:16

Hi, See this bug on github: https://github.com/broadinstitute/gatk/issues/5690

It has been 1 year and still not fixed.

It affects all tools that use smith waterman alignment from libgkl. Errors still occur on GATK 4.1.9.0.

 

Workaround for tools that have this option is using --smith-waterman JAVA.

==> However, FilterAlignmentArtifacts does not have the --smith-waterman option. It is hard coded to FASTEST_AVAILABLE: private static final SmithWatermanAligner ALIGNER = SmithWatermanAligner.getAligner(SmithWatermanAligner.Implementation.FASTEST_AVAILABLE);

This makes Mutect2 completely unusable on Terra/Google cloud VMs.

Can you please advise?

Thanks.

0

• 

  Bhanu Gandham

  • February 23, 2021 00:07

  Note: The solution for this query is being discussed in this github issue: https://github.com/broadinstitute/gatk/issues/5690

  0

  Comment actions Permalink
• 

  David Benjamin

  • March 03, 2021 04:14

  @GE Here is a PR to allow the Smith-Waterman implementation to be set in FilterAlignmentArtifacts: https://github.com/broadinstitute/gatk/pull/7105

  0

  Comment actions Permalink
• 

  GE

  • March 03, 2021 05:39

  Thank you

  0

  Comment actions Permalink

Please sign in to leave a comment.