search

broadinstitute / gatk-sv

A structural variation pipeline for short-read sequencing
BSD 3-Clause "New" or "Revised" License
162 stars 71 forks source link

WhamG memory issues hard to treat #624

Closed MattWellie closed 7 months ago

MattWellie commented 7 months ago

https://github.com/broadinstitute/gatk-sv/blob/dc92e9f4bc46a1ab19459b515c24c9747a73a967/wdl/Whamg.wdl#L220

This value is fixed at 0.0 and can't be edited. This is later used in a calculation, but it the result will always be 0.

https://github.com/broadinstitute/gatk-sv/blob/main/wdl/Whamg.wdl#L225-L229

We're running into some WhamG memory issues on some samples, but we can't bump the memory on a per-sample basis without handling pf_reads_improper_pairs & pct_exc_total; I've no concept of what reasonable values for these would be. The mem_per_bam_size * bam_size + mem_bam_offset calculation would be easier to influence, but due to the fixed 0.0 value it can't be edited.

mwalker174 commented 7 months ago

Hi @MattWellie,

You should be able to override the memory setting using runtime_attr_wham in GatherSampleEvidence, e.g. {'mem_gb': 16}.

MattWellie commented 7 months ago

Chained attribute override