Dev #1050
Dev #1050
Conversation
…te/seqr-loading-pipelines into benb/sv_gnomad_v4_migration
…s into benb/sv_gnomad_v4_migration
…s into benb/sv_gnomad_v4_migration
gnomad v4 sv migration
| ht = ht.annotate(gnomad_svs=sv.gnomad_svs(ht, gnomad_svs_ht)) | ||
| ht = ht.drop('info.GNOMAD_V4.1_TRUTH_VID') | ||
| return ht.annotate_globals( | ||
| versions=ht.globals.versions.annotate(gnomad_svs='1.0'), |
There was a problem hiding this comment.
wouldn't this already be the version- shouldn't we bump it?
There was a problem hiding this comment.
this migration will create the gnomad_svs table, but you're right, the globals annotation happens when we call ReferenceDataset.gnomad.get_ht so we don't need to set the version/enums here.
There was a problem hiding this comment.
oooh, there's some confusing bits here.
ReferenceDataset.gnomad_svs.get_ht sets the globals on the gnomad_svs reference dataset (gs://seqr-reference-data/v3.1/GRCh38/gnomad_svs/1.0.ht if persisted). This code was setting the globals on the SV annotations table as they're not currently there. It's not overwhelmingly important, as we exclude gnomad_svs from consideration when computing the reference datasets to update, and it would also be annotated by the normal process because I kept separate for_reference_genome_dataset_type_annotations (which includes gnomad_svs) and for_reference_genome_dataset_type_annotations_updates methods (which does not). I felt the migration should include the correct globals on annotations table for completeness though!
There was a problem hiding this comment.
right I got jumbled, this is an annotation table migration.
This code was setting the globals on the SV annotations table as they're not currently there
This is good reason to keep the globals annotation in this migration. @hanars, I'm going to ad the globals versions and enums annotations back and re-request review.
There was a problem hiding this comment.
yeah, it would be awesome if these had tests, but not entirely sure it would have caught this 😬.
| ht = ht.drop('info.GNOMAD_V4.1_TRUTH_VID') | ||
| return ht.annotate_globals( | ||
| versions=ht.globals.versions.annotate(gnomad_svs='1.0'), | ||
| enums=ht.globals.enums.annotate(gnomad_svs=hl.Struct()), |
There was a problem hiding this comment.
this feels unnecessary, as this would already be in the existing table globals and does not seem to be changing
There was a problem hiding this comment.
…e_allele_validation' into sv-locus-alleles
…tasettype_allele_validation bug fix: handle set of dataset types during allele type validation
No description provided.