[BioSQL-l] Re: [Bioperl-l] parent <-> subject etc

Aaron J Mackey ajm6q at virginia.edu
Tue Mar 25 07:21:42 EST 2003


On Tue, 25 Mar 2003, Hilmar Lapp wrote:

> <digression to="biosql">
> Quite honestly though, I'm not sure why your arguments wouldn't also
> apply to bioentry and seqfeature relationships - shouldn't we rename
> parent/child there too?
> </digression>

I think for seqfeatures, the parent/child stuff stems from the idea that
we were just going to store subseqfeatures; but as soon as you made that a
many-to-many sort of relationship, you invited the possibility of a graph,
not a strict hierarchical tree, and so now, yes, I agree with the above
"digression".  subject/object/predicate it is (with sub-seqfeatures
adhering to SO or some other ontology, perhaps the aforementioned internal
"biosql" ontology for API-supported stuff)

One ignorant question: someone mentioned that while parent<->child has an
inherent directionality to it; I also believe that subject<->object has a
similar directionality.  This isn't a bad thing, per se, but it does imply
that our graph structures are always directed (i.e. if I want to model
synonyms, I need relationships in both directions).  One should note that
"commutativity" in grammar is written with a "pair" as the subject: "These
two things are the same".  But in our data models all we get to do is "A
is the same as B; B is the same as A".  Is there some ontology-savvy
solution to this small conundrum?

Great discussion; I do wish we had it earlier.  There's now at least 4
ontology implementations floating about, all with strengths and
weaknesses.

-Aaron

-- 
 Aaron J Mackey
 Pearson Laboratory
 University of Virginia
 (434) 924-2821
 amackey at virginia.edu




More information about the BioSQL-l mailing list