Open ahwagner opened 10 months ago
I agree with this and think it merits further discussion. We should also revisit this in the context of switching over to LinkML, which was out of scope for our initial version of this work but might be a better fit long term. The LinkML representation seems well-suited to this use case, and is also gaining traction.
We currently provide a controlled vocabulary and should also include mappings from that vocabulary to concept codes where appropriate. We describe the use of multiple ontologies that are to be used: https://github.com/ave-dcd/mave_vocabulary/blob/96d3dcc0f59330ddd1d2ddcbf0b5ef9ff54a5cc4/README.md?plain=1#L101-L112
But our schema only provides the ability for codings to be used for
phenotypicAssay.relevance
(required; we recommend MONDO) and thephenotypicAssay.modelSystem.identifiers
(not required; we recommend the cell line ontology). We should update this section to either state that we provide a controlled vocabulary that maps to an ontology (and provide a SSSOM mapping; easy), or we update our schema to require the use ofcoding
objects (as we dophenotypicAssay.relevance
).