[Bioperl-l] Annotations and annotation tags can be Ontology Terms

[Hilmar Lapp]

On Tuesday, March 11, 2003, at 12:22  AM, Ewan Birney wrote:

> I like the ability to add GO and other ontology terms. Should we have 
> our
> own ontology of keywords, go terms etc etc.
>

While working on and with bioperl-db and biosql, I found myself wanting

	- an ontology of annotation tags ('dblink', 'keyword', 'comment', 
...), with the caveat that richly tagged datasources like UniGene and 
LL come with their own whole series of tags - it is useful though to 
have a machinery that tells you based on an ontology that 
'PREFERRED_OFFICIAL_GENE_NAME'-tagged annotation is-a 'gene_name' 
annotation, and

	- an ontology of bioentry-bioentry relationship types 
('has-transcript', 'has-translation', 'has-cluster-member', 
'has-family-member', 'is-homologous-to', 'has-synonym', etc)

>
> (another point - GO links could be represented as ontology term
> attachments or dblinks. You would prefer I guess ontology term
> attachments...)
>

Right. All the parsers right now don't utilize this though. The problem 
is that the datasources (e.g. Ensembl dumps :) will not tell you the 
difference between a synonym/translation/transcript/gene type of 
relationship behind a db_xref tag, or whether it is rather an ontology 
term annotation, like for GO terms. You'd have to interpret the ID to 
make a guess (not too hard for GO, but generally speaking you walk on 
unstable ground then).

	-hilmar
-- 
-------------------------------------------------------------
Hilmar Lapp                            email: lapp at gnf.org
GNF, San Diego, Ca. 92121              phone: +1-858-812-1757
-------------------------------------------------------------