Closed dpark01 closed 1 month ago
The merge of this to master
triggered a deployment to dnanexus, which is currently failing to compile via dxWDL:
https://github.com/broadinstitute/viral-pipelines/actions/runs/10237505135/job/28320935779
The issue seems to be with tasks_megablast.wdl
. The checks in this PR didn't catch it because deployment to dx is currently only triggered after a merge into master
. I'm looking into it and may change actions behavior so we actually test deployment to dx in PRs.
This PR:
assemble_denovo
on example exercises. This PR now has thescaffold
task fallback to old brute-force reference selection if ANI-based reference selection fails to find any matches at all. This will work fine (same as before) in most historically-normal use cases, but will not behave well if given a very large array of reference genomes to choose from. (this change does not impact the behavior ofscaffold_and_refine_multitaxa
which does not pass multiple references to thescaffold
task anyway.util.file.zstd_open
in favor ofzstandard.open
reports.alignment_metrics
samtools ampliconstats
in the case of multi-segment targets -- also allowsamtools ampliconstats
to fail silently in the case that it's getting too picky about the bed filemafft
workflow to usemulti_align_mafft_ref
instead ofmulti_align_mafft
terra_tsv_to_table
workflow resilient to non-existent TSVs in its input array, to simplify running it on a Terra table full of tsvs that may or may not exist