Add support for storing axiom annotations

[dosumis]

OBOish OWL ontologies store important information in axiom annotations. The majority of these are annotations on annotation axioms: Definitions are typically annotated with multiple xrefs to papers (mixed with references to the editors who have contributed edits). This is also frequently the case for synonyms.

VFB use of OLS/SOLR absolutely requires these annotation axoim annotations to be indexed. Where these use a standard reference ID (e.g. PMID), they also provide very useful information for users of OLS.

There is also increasing use of annotations on logical axioms in OBOish OWL ontologies (e.g. GO uses this to record whether a classification asserted in a pre-reasoned version of the ontology is supported by inference). VFB intends to use these in future for adding references and evidence codes to logical axioms. So, longer term, support for indexing annotation of logical axioms will be important.

[simonjupp]

OK, hadn't really thought about this for our SOLR index. I can dump these things into the index on a term for searching proposes, but you would probably lose the relation between the annotation and the annotation axiom so it would be no good for displaying these things. Currently for annotation we store a dynamic field annotation_* (e.g. annotation_definition) for all annotation on a term. We could easily add axiom_annotation_definition and lump them all in there, but if you had multiple definitions the mapping would be lost. We could add an additional field that stored the annotations + annotation axioms in a bit more structured form for display purposes. I'll get back to you on this.

[dosumis]

I can dump these things into the index on a term for searching proposes, but you would probably lose the relation between the annotation and the annotation axiom so it would be no good for displaying these things.

That would work OK for the definition xrefs, as there is only one def. But not for anything else. May be worth doing as we could live without displaying other axiom annotations, but not displaying this one would be a show-stopper.

[dosumis]

Hi again,

Wondering again about timeframe for this. Trying to work out whether to wait or write my own script to pull over missing content.

Cheers,
David

[simonjupp]

It's on my todo, and it's in the same area as #13 so hopefully I can get these done together.