Open paolaroncaglia opened 5 years ago
@cmungall We're aware of this work in MONDO on obsoletion of rare-X terms: https://github.com/monarch-initiative/mondo/issues/254 That ticket is closed. However, we see that there are still some 'rare disease' terms in MONDO, see my comment above. Could you please explain what strategy you followed - e.g. did you start from rare genetic diseases first, and are leaving other 'rare' terms for later? We'd like to align on this work, not last because any significant change in MONDO will result in EFO errors after your release :-) Many thanks in advance!
@paolaroncaglia To answer:
'rare' is already used in EFO3, but I assume that all its class usage cases in Protege (~900 of them) are for MONDO terms. @zoependlington could you confirm please?
I can confirm all uses of 'rare' are in MONDO terms or EFO terms that have been mapped to MONDO terms and therefore inherited the 'rare' from the MONDO mapping.
We obsoleted the ones for which there was a clear replacement, Rare X --> X, i.e. where we trivially shadow an existing class with a rare subclass
So ones like "rare disease with dentinogenesis imperfecta" remain, because we have no "disease with dentinogenesis imperfecta" (such a disease would already be rare I imagine)
I think we need to discuss these kinds of grouping classes (e.g. X with Y) separately. I think many could be obsoleted but this would be done in a very conservative way, in contrast to the Rare X --> X case.
On Wed, May 22, 2019 at 4:42 AM paolaroncaglia notifications@github.com wrote:
@cmungall https://github.com/cmungall We're aware of this work in MONDO on obsoletion of rare-X terms: monarch-initiative/mondo#254 https://github.com/monarch-initiative/mondo/issues/254 That ticket is closed. However, we see that there are still some 'rare disease' terms in MONDO, see my comment above. Could you please explain what strategy you followed - e.g. did you start from rare genetic diseases first, and are leaving other 'rare' terms for later? We'd like to align on this work, not last because any significant change in MONDO will result in EFO errors after your release :-) Many thanks in advance!
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/EBISPOT/efo/issues/447?email_source=notifications&email_token=AAAMMOPJ3NAYYKFESSV3ALLPWUWTBA5CNFSM4HNK2E32YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODV6YZIQ#issuecomment-494767266, or mute the thread https://github.com/notifications/unsubscribe-auth/AAAMMOO5R7WLE4Q3WKQ3FVTPWUWTBANCNFSM4HNK2E3Q .
Paola, can we compile a list of all the Rare X -> X that would be candidates for removal from EFO? We should check internally if any of these have been used for direct annotation and ensure there's a migration path for the downstream consumers.
@cmungall thanks for explaining. @simonjupp I opened a ticket for that task, assigned it to me and linked it to this epic.
Update on obsoleting (some) Orphanet 'rare x' diseases in EFO: part of this task will overlap with cleaning up undesired Orphanet-derived superclasses (Epic https://github.com/EBISPOT/efo/issues/322). That work is now scheduled for the first quarter of 2020 as it's top priority for Open targets.
Update: we are still working on this in collaboration with Mondo; might be addressed in the second quarter of 2020 as part of the Orphanet work.
Update: we're working with Mondo to address 'rare disease' terms that are still in Mondo and that Mondo already tagged as obsoletion candidates. Those are Orphanet-derived terms, so addressing them also addresses some Orphanet-derived parentage issues. In detail, Nicole and I have been working on this spreadsheet. Making a note here to
@zoependlington I haven't worked on disease terms since the time of my latest comment, but I know that Mondo has progressed steadily with its cleanup of 'rare disease' terms. The results of their work-in-progress should then all percolate to EFO eventually. Is this ticket still needed, and if yes, should it be updated?
Yes, any remaining rare terms should be Orphanet terms that will eventually be obsoleted/replaced by Mondo terms. There may be a few EFO terms that need looking into however, so this ticket should remain open.
Background: EFO agrees that representation of rare diseases is unsatisfactory; MONDO has already taken steps to rectify the situation (https://github.com/monarch-initiative/mondo/issues/254). EFO may want to adopt similar solutions (or align to MONDO altogether, to minimize inconsistencies that need time and effort to fix in EFO3). Open Targets already showed interest in an advanced solution (generate disease prevalence statements, #9).
To be sorted: