[BioSQL-l] Re: Ontology.category

[Chris Mungall]

On Sun, 22 Sep 2002, Hilmar Lapp wrote:

> On bioperl we're currently thrashing out the best interface for
> Bio::Ontology::Term. ChrisM proposes a link from Term to the
> 'ontology category', i.e. a virtual root term or something. I think
> having this is very useful; Dave Block has just hit this problem.
>
> Of course it can be solved in the db by a materialized view that
> multiplies all relationships, and then you check for that parent
> that doesn't have parents. Somewhat tedious though. Also, foregoes
> the chance to have a UK on name within an ontology. Since I strongly
> believe that there should be a UK on name, within an ontology or
> globally (in fact, I've put one there), this means that with the

not globally - in GO, there are two distinct terms 'glycine-gated chloride
channel', one pertaining to function the other to cellular component

> present design terms can (often will) live in (have a path to) more
> than one ontology. The problem with this is that the definition of a
> term might not be exactly the same between different ontologies,
> i.e., the semantics of a particular term may differ depending on
> what the context-ontology is. You cannot capture this in the present
> design.
>
> The alternative is to add a FK category_id on Ontology_Term to
> itself that shall point to the term specifying the name of the
> ontology. Subsequently, one can put a UK on (term_name,category_id).

perfect

> Thoughts/comments/preferences/votes?
>
> 	-hilmar
> --
> -------------------------------------------------------------
> Hilmar Lapp                            email: lapp at gnf.org
> GNF, San Diego, Ca. 92121              phone: +1-858-812-1757
> -------------------------------------------------------------
>
>