effects "all" does not catch empty annotations

[stolpeo]

Setting the effects to catch all annotations (all) (also used by preset impact to any) does not include variants that have no annotation at all. Users expect to catch all variants when they loosen the settings as much as possible. The postgres query looks like this;

variants_smallvariant.refseq_effect && cast(array[
   "3_prime_UTR_exon_variant",
   "3_prime_UTR_intron_variant",
   "5_prime_UTR_exon_variant",
   "5_prime_UTR_intron_variant",
   "coding_transcript_intron_variant",
   "complex_substitution",
   "direct_tandem_duplication",
   "disruptive_inframe_deletion",
   "disruptive_inframe_insertion",
   "downstream_gene_variant",
   "exon_loss_variant",
   "feature_truncation",
   "frameshift_elongation",
   "frameshift_truncation",
   "frameshift_variant",
   "inframe_deletion",
   "inframe_insertion",
   "intergenic_variant",
   "internal_feature_elongation",
   "missense_variant",
   "mnv",
   "non_coding_transcript_exon_variant",
   "non_coding_transcript_intron_variant",
   "splice_acceptor_variant",
   "splice_donor_variant",
   "splice_region_variant",
   "start_lost",
   "stop_gained",
   "stop_lost",
   "stop_retained_variant",
   "structural_variant",
   "synonymous_variant",
   "transcript_ablation",
   "upstream_gene_variant"
]::varchar[] AS varchar[])

It should also be valid if no annotation is available in case all is set.

[stolpeo]

Not sure if we really want this or if this would cause to much artifacts. There could also be just a checkbox to disable the effect filter in the query so it shows even not annotated variants. I guess this is up for discussion.

[xiamaz]

If users select all they might have very small regions of interest. I think this should be ok. Too many artifacts should be caught by Quality filters, not by this one.

[stolpeo]

If the question comes up, disabling all effects shows no variants (as expected) variants_smallvariant.refseq_effect && cast('{}'::varchar[] AS varchar[]), because an entry does not overlap with anything in an empty list.

[xiamaz]

@holtgrewe Should every variant in mehari theoretically have an annotation or is no annotation sane behavior?

[xiamaz]

@stolpeo In default behavior "all" should also return unannotated variants. In this specific instance we also have annotation issues, so there might also be some resulting mehari changes.

[stolpeo]

OK, could you prioritize this ticket? @xiamaz

[xiamaz]

Prioritized. If change is small please handle soon, as this might allow users to work around potentially significant mehari annotation issues.