nonhuman genetic disease pattern #7794
Conversation
Addresses #6085
| - gene | ||
|
|
||
| equivalentTo: | ||
| text: '%s and ''has material basis in germline mutation in'' some %s' |
There was a problem hiding this comment.
You need either
- a property chain
'has material basis in germline mutation in' o 'in taxon' -> 'in taxon', in RO - the ensure the genus is already species-specific. Presumably this is not always the case as some PS analogs will be cross-mammal, cross-vertebrate, etc
- just make a var for taxon even if it is duplicative with either/both genus/differentia
But going back to your Q from slack about infectious disease classification:
- I would in general try and keep the ladders parallel. If some order/class/etc level virus is a useful grouping for H, it should also be for NH? But I appreciate this may not always be the case, esp for agents that never infect human, in which case
- create a
non_human_disease_by_infectious_agentDP- this would only use NH terms as genus
- the taxon of the host would need to go in as a variable
- classification would follow NCBITaxonomy
- we can worry about making the xp-analog relations complete via rules later
If you do go with 2 it points to doing the same this for this PR, e.g. including a taxon var
|
Best add to Tech call agenda |
|
Added, but the next Tech call is not until Friday, June 21 |
|
@katiermullen @matentzn @cmungall |
There was a problem hiding this comment.
@sabrinatoro I have one tiny comment. Can you please take a look? Otherwise, it looks great to me and can be merged.
| @@ -22,38 +24,45 @@ annotationProperties: | |||
| vars: | |||
| disease: '''disease''' | |||
| gene: '''gene''' | |||
| taxon: "'taxon'" | |||
There was a problem hiding this comment.
For my education - is this supposed to be '''taxon''' and not "'taxon'"?
There was a problem hiding this comment.
Probably both are ok.. but definetly always the same. Good eyes!
|
|
||
| classes: | ||
| disease: MONDO:0005583 | ||
| gene: SO:0001217 |
There was a problem hiding this comment.
This is correct but maybe an unnecessary complication for the matching system. Consider using SO:0000704 instead so that DOSDP match always works even if the source does not distinguish between protein codeing and not protein coding genes.
There was a problem hiding this comment.
Good call. I will also update the human pattern
| @@ -22,38 +24,45 @@ annotationProperties: | |||
| vars: | |||
| disease: '''disease''' | |||
| gene: '''gene''' | |||
| taxon: "'taxon'" | |||
There was a problem hiding this comment.
Probably both are ok.. but definetly always the same. Good eyes!
| - gene | ||
|
|
||
| equivalentTo: | ||
| text: '%s and (''has material basis in germline mutation in'' some %s) and (''in taxon'' some %s)' |
There was a problem hiding this comment.
Note here that there are two ways you can model this:
- you can push the taxon into the gene. assuming you always use hgnc identifiers for humans and ncbigenes for animals, and you always use the correct gene id pertaining to a taxon specific gene, this is enough, no need for recording the taxon specifically to distinguish from human disease
- however, you likely want the taxon to separate grouping classes "dog disease", which means having the taxon can be helpul (unless the genes in our ingest have them already!). You might also need this for non-gene non-human-animal diseases, so maybe just curate redundantly. If you do, however leave it like you have it, not that technically, you may want to add the human taxon to the human diseases.
Long story short: its probably best to leave it has you have it, just adding some considerations in for future decision making.
There was a problem hiding this comment.
After a lot of discussion, we decided to keep the redundancy. But I agree, we might decide to remove this redundancy at a later time
Addresses #6085
I also updated the existing patterns for genetic diseases in human to point to the specific "human disease" and not general "disease"