search

clingen-data-model / clingen-interpretation

Allele (variant) interpretation model and API for ClinGen
3 stars 1 forks source link

Display Names in Value Sets #151

Closed cbizon closed 6 years ago

cbizon commented 6 years ago

@bpow @larrybabb

The main thing I need to sort out for json_interpretation is the changes away from Coding/Codeable concept.

When we get stuff from the VCI, it does not have ID values. So it says "Likely Pathogenic", not "LN:LA26333-7". In the days of codings, we had those things together, so I wrote some code so that we could translate easily. But now, that translation has been lost, because our value sets don't contain the term names.

Could we re-attach the names somewhere without making them a necessary component of Domain Entity? Otherwise, I don't know how to transform VCI data without hard coding a lot of crap in there...

bpow commented 6 years ago

With the commit referenced above, the VSxxx files now reify the IdentifierSystemConcepts to include "display", which I think should give the textual description you need for the translator.

@cbizon and I also discussed whether there should be a more formal mechanism for translating from VCI terms to our terminology (maybe by mapping through a JSON-LD context), but we might do something more "hard-coded" for the short term.

cbizon commented 6 years ago

OK, this is good. I'm going to leave the ticket open for the moment so that we can discuss the long-term solutions more fully. The suggestion that Bradford had was when we get something from VCI like

"strength": "moderate"

that we have a context that says that "moderate": "LN:LA1234" or whatever.

The question is whether this kind of IRI expansion/interpolation happens in values or just in keywords. THere's some stuff in the spec about Typed Values, but I found it a bit incomprehensible at first pass.