Allow flexible variant input

[brettpthomas]

Right now we accept RSID for variant input. We should allow clinicians to enter F508del, among other things

[brettpthomas]

Andrew and I just chatted about this. Problem: most HGVS notations we've seen are tied to a refseq transcript, not ensembl. So setting up an HGVS -> VCF mapping within the browser would be hard.

This is a different problem than the Leiden parsing. I don't mind if we only support a subset of HGVS notations, but any that we do support must be 100% accurate. I don't know how to do that without storing (versioned) refseq transcripts in the browser alongside ensembl ones, which seems like a lot of overhead for this feature.

So - leaving this open and let's discuss next time we discuss browser functionality as a group

[konradjk]

Ah, makes sense. I feel like I've seen them with gene symbol, but maybe that's mostly colloquial (and made a bit difficult due to versioning).

[andrewhill157]

We did discuss allowing a gene symbol and then picking the canonical ensembl transcript as the reference. As Brett mentioned, versioning and refSeq vs. ensembl mismatches could cause headaches. If I had to guess, most HGVS coordinates are relative to a specific refseq transcript version, even though clinicians don't always explicitly state that is the case (all LOVD entries were in this format).