search

PHI-base / phipo

Pathogen-Host Interaction Phenotype Ontology
Other
5 stars 5 forks source link

Should the AE terms in 'infective ability' and 'outcome of interaction phenotype' be PHIPO terms or PHIPO_EXT terms? #354

Closed CuzickA closed 2 years ago

CuzickA commented 2 years ago

image

These terms don't really describe the 'observed phenotypes' that we are capturing with PHIPO terms. They are more conclusion outcome terms which is why we have added them as Annotation Extensions.

Currently these terms are findable in PHIPO and may be used by others in the future. We have a small comment in the text about only using them for AE in PHI-Canto but this is easily missed (and currently absent in the outcome of interaction child terms). Do we want other users to use these ontology terms or should we change them into PHIPO_EXT terms only to be used in PHI-Canto in the same way as we have dealt with the Gene-for-Gene PHIPO_EXT terms?

egs https://www.ebi.ac.uk/ols/ontologies/phipo/terms?iri=http%3A%2F%2Fpurl.obolibrary.org%2Fobo%2FPHIPO_0001179 https://www.ebi.ac.uk/ols/ontologies/phipo/terms?iri=http%3A%2F%2Fpurl.obolibrary.org%2Fobo%2FPHIPO_0000015 https://www.ebi.ac.uk/ols/search?q=disease+present

If we do this will it have much of a knock-on effect about how we display the data in the new MC PHI-base 5/6 interfaces?

We would still need a method of showing the term definitions to the PHI-base data users.

jseager7 commented 2 years ago

Do we want other users to use these ontology terms or should we change them into PHIPO_EXT terms only to be used in PHI-Canto in the same way as we have dealt with the Gene-for-Gene PHIPO_EXT terms?

One argument for keeping these terms in PHIPO proper is because they're effectively the key distinguishing feature of PHIPO compared to other phenotype ontologies. Pathogenicity and virulence are concepts intimately related to pathogen-host interactions, so users might be surprised not to see those in the ontology.

Also, if uPheno ends up adding patterns for pathogenicity and virulence, then that's a strong argument for keeping the terms in PHIPO (admittedly the patterns could be applied to terms in PHIPO_EXT, but we'll likely never get PHIPO_EXT registered on the OBO Foundry so its use is always going to be limited).

I also don't think it's fair on other ontology users for us to insist that others use the ontology in a particular way. It's fine for us to restrict usage in PHI-Canto (we already do), but I think we should try to make PHIPO as generally applicable as possible in case it gets picked up by other pathology groups in future. As mentioned above, they'll probably want terms describing changes in pathogenicity and virulence (assuming they share our understanding of them).

jseager7 commented 2 years ago

If we do this will it have much of a knock-on effect about how we display the data in the new MC PHI-base 5/6 interfaces?

I don't think the knock-on effect in terms of display will be a problem, because we're just moving data around: PHI-base 5 can display extensions and phenotypes already and I would very much hope that PHI-base 6 will handle new annotation extensions in the same way.

We would still need a method of showing the term definitions to the PHI-base data users.

The PHI-base 5 beta already displays definitions for annotation extension terms, so I don't think any updates would be needed here: image

CuzickA commented 2 years ago

@jseager7 I agree. I think we discussed this in the past and couldn't remember our decision. Let's leave them in PHIPO for now. Closing ticket.