biolink-model
biolink-model copied to clipboard
Promote GenomicEntity to be a concrete class instead of a mixin
Is your feature request related to a problem? Please describe. On NCATS Translator https://github.com/NCATSTranslator/Feedback/issues/586, we've been trying to figure out what Biolink class to map UMLS semantic type T028 "Gene or Genome" to. It is currently mapped to mapped to the mixin biolink:GenomicEntity, but since biolink:GenomicEntity is not a concrete class, we need some additional information to fully map a UMLS concept like UMLS:C3890397 "PRKAA1 wt Allele", which as its name suggests is an allele. But since its only semantic type is T028, we don't have any other information on what Biolink type this entity should have, so we need to decide between removing it entirely or manually assigning it to another Biolink type.
What working group (or team) did this request originate from? NCATS Translator Node Normalizer
Note: This is relevant for members of NCATS Translator.
Describe the solution you'd like Making biolink:GenomicEntity a concrete class would simplify this tremendously. It could be a child of biolink:BiologicalEntity and could be a parent of biolink:Gene and biolink:Genome. I understand that this would be a pretty major change, so we're happy to consider any simpler solutions you might have.
(I may also be misunderstanding how Biolink works -- if biolink:GenomicEntity already has a parent class that I can't access from bmt for some reason, please let me know and I'll look into doing that!)
Additional information to support this request (optional)
- Previous (and ongoing?) discussion about mapping STY:T028 at https://github.com/biolink/biolink-model/issues/1008
Tag relevant members for discussion @sierra-moxon @mbrush @cbizon @saramsey
Thank you @gaurav for logging this issue.
Our team (Expander Agent) supports making GenomicEntity
a concrete (i.e., non-mixin) class. It would give us options for automatically assigning a Biolink category for UMLS concepts with semantic type T028
. We cannot manually curate each of these concepts; there has to be an allowed mapping for it, that can be applied automatically. Using GenomicEntity
for this purpose seems like a simple practical solution.
@nlharris Just wanted to poke y'all about this ticket. If you need any more information from us, please let us know!
Hi @gaurav, I'm not sure what the status is. Perhaps @sierra-moxon can comment?
I think it could work - I am a bit worried about nucleic acid entity
(and children) which, without any other changes, will no longer be related to the genomic entity
hierarchy. But gene and genome will be, so it could be ok.
FWIW, I think it's confusing that NucleicAcidEntity
is subsumed by GenomicEntity
; I think it should be the other way around, as synthetic nucleic acid entities could be non-genomically based, but I think that's being glossed over. Additionally, NucleicAcidEntity currently seems to be defined essentially as representing genomic entities to me...