search

broadinstitute / genetic-prevalence-estimator

https://genie.broadinstitute.org/
BSD 3-Clause "New" or "Revised" License
1 stars 0 forks source link

Allow specifying a transcript for custom variant lists #74

Closed sambaxter closed 2 years ago

sambaxter commented 2 years ago

The wrong variant details (p./c.) are showing up that are for a different gene in the variant list. I think this case because it is an odd issue because there is a pathogenic intronic variant in our gene of interest, but it is also located in another gene. Since the user specifies the gene and transcript, can we make sure the variant name always uses the detail from that transcript? Maybe we should also add gene to the table just so people can pick this out more easily, just in case it does happen in other. random situations? image

nawatts commented 2 years ago

Was this a recommended or custom variant list? Currently, we only have gene and transcript specified for recommended variant lists. Custom variant lists behave similarly to the gnomAD browser's region page and show the most severe consequence in any transcript. https://github.com/broadinstitute/aggregate-frequency-calculator/blob/ea6df3cf7ed44564b2a3055229d8a6389c6ead93/worker/src/worker/tasks.py#L304

sambaxter commented 2 years ago

This variant came across in both a custom list (https://website-sauthoia2q-uc.a.run.app/variant-lists/cf6aab40-39a8-493f-ab79-cc6ff7487d43/) and in a recommended list (https://website-sauthoia2q-uc.a.run.app/variant-lists/35eae945-5185-437b-a2db-4f836ac17390/) and it comes across with the right nomenclature in the recommended list. Makes sense for the custom list that it wouldn't know the gene or transcript. Is that something we could have them put in but not require? If not, its not a problem. I will just write something about it the documentation

nawatts commented 2 years ago

Makes sense for the custom list that it wouldn't know the gene or transcript. Is that something we could have them put in but not require? If not, its not a problem.

Yes, we could do that. A few follow up questions...

If a gene/transcript is not provided, does showing the most severe consequence make sense? Or should we remove the VEP consequence and HGVS columns from the table in that case?

What to do if a gene/transcript is provided and the list contains variants where gnomAD doesn't have a VEP consequence in the given gene/transcript for the variant? I'll suggest maybe showing a flag in the VEP/HGVS columns (similar to what we do for AC = 0 variants).

nawatts commented 2 years ago

I guess that, if we're going to be showing the most severe consequence of a variant in the table, we should add a column to the table showing which gene/transcript that consequence occurs in.

sambaxter commented 2 years ago

If a gene/transcript is not provided, does showing the most severe consequence make sense? Or should we remove the VEP consequence and HGVS columns from the table in that case? Yes, if they don't provide gene/transcript we would put the most severe consequence

What to do if a gene/transcript is provided and the list contains variants where gnomAD doesn't have a VEP consequence in the given gene/transcript for the variant? I'll suggest maybe showing a flag in the VEP/HGVS columns (similar to what we do for AC = 0 variants). I think your idea about the flag is the right choice

I guess that, if we're going to be showing the most severe consequence of a variant in the table, we should add a column to the table showing which gene/transcript that consequence occurs in. Yes, I think adding the gene/transcript to the table makes sense. Think its worth adding for both the recommended and custom lists.