search

tonydisera / gene.iobio

An iobio app for examining gene variants
22 stars 4 forks source link

ClinVar badge showing in gene button, but not in Ranked Variants due to filtering to only show PASS VCF filter status. #286

Closed tonydisera closed 8 years ago

tonydisera commented 8 years ago

I was a bit perplexed to see a ClinVar glyph in the gene button, but no corresponding variant in the Ranked Variants card (or variants card).

screen shot 2016-03-22 at 10 00 24 am

But when I unclick the "PASS" filter for VCF filter status, the variant appears:

screen shot 2016-03-22 at 10 00 32 am

Since the VCF filter status is excluding non-pass records, the code that looks at the variants to determine the highest impact ClinVar (and other categories, like SIFT, PolyPhen, etc), should exclude these same records by default.

Al, what do you think? I guess the downside is that we run the risk of not bringing attention to a variant that is highly pathogenic, yet doesn't pass VCF quality checks.

AlistairNWard commented 8 years ago

Good question. I think it is valuable to see that there is evidence for a pathogenic variant, even though it is filtered out. We should maybe discuss this on Skype and get Chase in as well? My initial thoughts are:

  1. We could introduce new panel on the left called 'Gene info' or something. When you move to a new gene, this panel opens up by default. In this panel, we could include info on the gene (like what you see hovering over the gene badge), but also info on things going on in the gene. So we could include in this panel that there are filtered variants that are filtered out and the filters that are causing them to not be present. This will come up again when we allow default filters to be automatically applied. For example, if, by default a depth > 20 filter is applied, but a variant with DP=19 is pathogenic in ClinVar, we would want the user alerted. So maybe this gene level panel would do the job?
  2. Anything that is marked up in some interesting way (e.g. pathogenic in Clinvar, SIFT etc) should appear in the ranked variants table even though the variant is filtered out and not visible. The column for that variant would be outlined or something to highlight that it is filtered out?
tonydisera commented 8 years ago

Hmm... I favor option #2 because it is simpler, but hesitate to have the variants in the ranked variants card not match the variants shown in the variants/sample card.

This is a similar issue to the alternate transcript. We want to alert the user that there is a high-impact variant, but it is not in the current visible set of variants. Perhaps we should address both issues with one design.

AlistairNWard commented 8 years ago

382 is handling this now.