Closed cmungall closed 6 months ago
random effect model
(('has part' min 1 'model error term') and ('has part' min 1 ('model random effect term' and ('is about' some 'study design independent variable')))) and ('is about' some 'study design dependent variable')
IMO this is a rococo axiom and oculd be removed
I came across this issue when building a STATO import module using the Ontology Development Kit, which has OWL 2 DL as the default profile. As I do not want to change the expressivity profile of my ontology project, I'd second the proposal by @cmungall to drop or at least weaken the above mentioned axioms.
Unfortunately, the build pipeline of STATO is not well enough documented to allow a drive-by PR. It seems that the edit file is the /dev/ontology/stato.owl. If this is correct, I suggest to do the following edits on it, to get rid of all the "non-simple property in a restriction" errors:
'mixed effect model' (STATO:0000189)
drop the subclassOf axiom 'has part' min 1 ('fixed effect model' and 'random effect model')
or relax it to 'has part' some ('fixed effect model' and 'random effect model')
'random effect model' (STATO:0000099)
drop (('has part' min 1 'model error term') and ('has part' min 1 ('model random effect term' and ('is about' some 'study design independent variable')))) and ('is about' some 'study design dependent variable')
or relax it to (('has part' some 'model error term') and ('has part' some ('model random effect term' and ('is about' some 'study design independent variable')))) and ('is about' some 'study design dependent variable')
'fixed effect model' (STATO:0000082)
drop ('has part' some ('model interaction effect term' and ('is about' exactly 2 'study design independent variable'))) and ('has part' min 1 'model error term') and ('has part' min 1 ('model fixed effect term' and ('is about' some 'study design independent variable')))
or relax it to ('has part' some ('model interaction effect term' and ('is about' some 'study design independent variable'))) and ('has part' some 'model error term') and ('has part' some ('model fixed effect term' and ('is about' some 'study design independent variable')))
Regarding the "reserved vocabulary for annotation property IRI" error, I have not found where this axiom comes from. It is in none of the owl files that live in the dev folder. So I assume it was added manually to the release files living in 1.4 and latest_release. But to fix it one would have to delete the axiom annotation highlighted in the below screen shot and then delete the rdf:type annotation property declaration.
@cmungall @StroemPhi working on it