Term request - frequency_in_condition

[kshefchek]

It would be useful to have a relationship to link an instance of a phenotype to the frequency seen in a cohort of patients diagnosed with a condition. This could be either an object property (eg subClasses of https://hpo.jax.org/app/browse/term/HP:0040279), or a datatype property.

[cmungall]

One way to model this is using a basic population size property. E.g. pop1 is the cohort, pop2 is the subset of the cohort with the phenotype. If you know the size of each you have the frequency.

This would obviously be very verbose so I think having a shortcut property would be useful here. I think there is also a use case for attaching the property to the cohort->phenotype edge, or more generally to any edge.

E.g.

?s ?p ?o << ro:subject_has_object_with_frequency 0.5>>

(see this post ).

[kshefchek]

I prefer the latter but can work with either. I think in most cases of phenotype-disease from the HPO this is a qualitative value.

[dosumis]

A generic way to model phenotype penetrance would be v.useful. This is tricky to do well in OWL if we want to be able to roll penetrance into classes. For lethality modelling I used a reified data property 'mortality rate' as, for the pattern I used, a generic penetrance data property => problems with co-reference resolution. For annotation I prefer the edge tagging approach as easy to use this for filtering edges. Can we call this has_penetrance? It would also be useful to have a way to make qualitative assertions of penetrance (low, medium, high). Makes sense to me to use edge tagging for this too.

[mah11]

In PomBase FYPO phenotype annotations, penetrance is one of the things for which we use (traditional, pre-GO-CAM) GO-style annotation extensions. The relation is indeed has_penetrance, and the range accepts numerical values (as percents) or qualitative values selected from a tiny in-house CV (low, medium, high, full/complete, variable).

I'm not familiar with the "edge tagging" jargon, so I don't know if our use of extensions is the same thing or similar.

If I understand "roll penetrance into classes" correctly, we don't want to do it. We want penetrance associated with annotations, and not by using different ontology classes for different penetrance observations.

(Minor aside: it would not have occurred to me to use "frequency" to refer to incidence or penetrance, but that may be because I've spent so many years steeped in single-celled organisms ...)

[kshefchek]

I'm not sure we should conflate penetrance with disease phenotype frequency from the HPO, although these might be the same value in mendelian disorders. Perhaps two new properties are needed?

[cmungall]

Yes, might need a hierarchy of data props.

Can we have a more descriptive name 'has genetic penetrance' - ugly I know but RO has a broad community not just geneticists...

On Thu, Aug 15, 2019 at 10:12 AM Kent Shefchek [email protected] wrote:

I'm not sure we should conflate penetrance with disease phenotype frequency from the HPO, although these might be the same value in mendelian disorders. Perhaps two new properties are needed?

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/oborel/obo-relations/issues/332?email_source=notifications&email_token=AAAMMOPNNX2Y6Y5GGE6EPILQEWE7NA5CNFSM4ILPQMA2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOD4MNGTA#issuecomment-521720652, or mute the thread https://github.com/notifications/unsubscribe-auth/AAAMMOOGOTNA4VS3JRG2QADQEWE7NANCNFSM4ILPQMAQ .

[kshefchek]

checking if this is on anyone's radar, it would be good to get rid of our fake IRI's in monarch

[dosumis]

checking if this is anyone's radar, it would be good to get rid of our fake IRI's in monarch

Looking over this again, I think it makes sense to have a generic way to record frequency in a population by tagging phenotype assertions (given that penetrance apparently doesn’t work for non-mendelian traits (?)). This could be via direct tagging of edges with a has_frequency AP, but if we need to be able to record something about cohort/population for which frequency is recorded, then perhaps better to attach an OWL:individual to edge and attach assertions about frequency and population to that individual - something like we do for provenance. Or - I guess we could allow for both patterns.

[nlharris]

What is the status of this?

[github-actions[bot]]

This issue has not seen any activity in the past 2 years. It will be closed automatically 60 days from now if no action is taken.

[github-actions[bot]]

This issue has been closed automatically because it has not been updated in over two years. Please re-open if you still need this to be addressed.