two questions: DAML+OIL, mapping to KEGG 'orthology'

[Chris Wroe]

Paul,

At 15:19 15/05/2002 -0700, you wrote:
>On the recommendation of Midori Harris, I wish to ask
>a couple of questions:
>
>First:  I have seen some mention of presenting the GO ontologies
>in DAML+OIL.  Has this been done?  If so, are you pleased with
>the expressiveness of this approach, and would you recommend it?


I'm involved in the Gene Ontology Next Generation (GONG) Project whose 
major aim is to evaluate whether the expressiveness and reasoning 
capabilities of DAML+OIL will support the future development, maintenance 
and use of GO.

There is a minimal web page at http://img.cs.man.ac.uk/gong

We have translated GO RDF to DAML+OIL but this is purely a syntactic 
transformation.
The next step is to 'dissect' the definitions of a sample of the GO terms 
into a property based DAML+OIL description.

So for example:

GO:0004054 arginine kinase may be (for sake of argument) be described as
         enzyme which [catalyses and only catalyses] (phosphorylation which 
[acts_on and only acts_on] L-arginine)

GO:0019202 amino acid kinase defined as
         enzyme which [catalyses and only catalyses] (phosphorylation which 
[acts_on and only acts_on] amino acid)

(These aren't complete definitions as they don't talk about the other 
substrate and product but anyway I can send you the DAML+OIL for this example.)

When we have a classification of small molecules including amino acids the 
DAML+OIL reasoner will be able to infer arginine kinase is-a amino acid 
kinase. It will flag as inconsistent an attempt to manually assert 
fructokinase is-a amino acid kinase. Finally, it opens up the possibility 
of  dynamically composing concepts as required (analogous to GO cross 
products) formally constraining what is reasonable to assemble together.

It is early days in the GONG project. DAML+OIL does limit its 
expressiveness to allow decidable reasoning. However, as long as you limit 
the use of DAML+OIL to terminological information (for which it was 
intended) then it is fine. I am actually used to using a less expressive DL 
based ontology language (GRAIL).

Actually the biggest barrier to the use of DAML+OIL and similar DL based 
ontologies is the added complexity they introduce. A major focus of the 
GONG project is to build an environment to make migration to the added 
expressiveness of DAML+OIL as simple as possible and to simplify the 
deployment of the resulting DAML+OIL ontology.

Chris


>Second: Do you know of any attempt to map between (or even
>unify) the GO ontologies, and the 'orthology' used by KEGG?
>Here at the Institute of Systems Biology, our biologists are
>very fond of KEGG; I -- as a programmer -- am naturally
>drawn to clean design of GO, which I have used for the last year.
>
>I want to make both KEGG and GO functional classifications
>available here, and I want to use data structures and
>ontology representations which make the most sense.  If you
>have any advice for me, I'd love to hear it.
>
>Many thanks,
>
>   - Paul Shannon
>
>--
>This message is from the GOFriends moderated mailing list.  A list of public
>announcements and discussion of the Gene Ontology (GO) project.
>Problems with the list?           E-mail: owner-gofriends at geneontology.org
>Subscribing   send   "subscribe"   to   gofriends-request at geneontology.org
>Unsubscribing send   "unsubscribe"  to  gofriends-request at geneontology.org
>Web:          http://www.geneontology.org/

Dr Chris Wroe
Clinical Research Fellow
Department of Computer Science
University of Manchester
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.stanford.edu/pipermail/go-friends/attachments/20020517/52d0a350/attachment.html>