broadinstitute / seqr

web-based analysis tool for rare disease genomics
GNU Affero General Public License v3.0
175 stars 89 forks source link

Including high SpliceAI score as a type of always returned (like ClinVar variants) #1139

Closed anneodonnell closed 3 years ago

anneodonnell commented 4 years ago

Letting high SpliceAI scores (maybe check boxes for red and yellow range) where variants with high spliceAI scores will be returned like ClinVar variants get returned (following same rules as ClinVar).

hanars commented 4 years ago

We also have this open ticket about filtering by cadd: https://github.com/macarthur-lab/seqr/issues/361

I'll probably want to implement all the in silico filtering in one coherent update. Could you fill in some details here about exactly which in silico predictors we want to be able to filter by, and how we want overriding to work?

bw2 commented 4 years ago

Personally, I found it intuitive to have a single Annotations section like image

I think that way it's clearer that selecting more checkboxes leads to more search results, and these SpliceAI and other filters would fit there.

hanars commented 4 years ago

I would want to lay these out a bit differently, but I'm not totally opposed to having them in one big section. That said, I would really want actual analyst feedbac before making that decision. I personally find it really confusing to display things as though being clinvar pathogenic is in any way the same as being a frameshift variant is anything the same as having a high splice AI

bw2 commented 4 years ago

they are technically all annotations and the unifying section logic is "show any variant that has/is at least one of these".

hanars commented 4 years ago

Like I said, you and I feel differently and the opinion I care about is the analysts'

hanars commented 3 years ago

Duplicated by https://github.com/broadinstitute/seqr/issues/2186