[Biojava-l] Ontology OBO is_a TERMs

[Andreas Prlic]

I am the one who added the OboFileHandler (based on some original code
from obo-edit). I was not sure how best to build up the datastructure
representing the relationships in a memory efficient way at that time,
so I left it out . Does anybody already have a solution from another
project for that, that we could use here?  Both links and triples
sound like reasonable approaches to me. I think the original ideas in
the Ontology framework were to support triples. I can have a look how
difficult it would be to build up the hierarchy using that.  (after I
added some feature requests for the structure modules...)

Andreas






On Thu, Jul 23, 2009 at 8:39 AM, Peter Midford<peter.midford at gmail.com> wrote:
> JP,
>      No, these are not just annotations to terms, and the code jumbles
> together several things that should be separated.  To properly handle the
> IS_A key, you will have to build the hierarchy, which you can do OBO style
> using links or restrictions (which I believe are a subclass of links in OBO,
> rather than adding an intermediate class to the ontology) or OWL style using
> triples (Subject, Predicate, Object) where is_a would be your predicate.  I
> assume the other key values that look like set operations are for building
> restrictions.
>
> Peter
>
>
> On Jul 23, 2009, at 11:29, JP wrote:
>
>> I never quite got this Peter, what is a triple ?
>> Could these simply be considered as annotations ?  Or are you thinking in
>> the lines of building hierarchies out of these (I take it this is the most
>> common task).
>>
>> These relationships are *fundamental* for any work of ontology.
>>
>> 2009/7/23 Peter Midford <peter.midford at gmail.com>
>> JP,
>>       Looking at the code for OboFileHandler.java, fresh from svn, it
>> looks like they're presently being dropped on the floor.  Perhaps someone
>> should either implement obo restrictions or links or build some triples
>> here, as they seem to be used for the rest of the ontology code.
>>
>>
>>
>>                        } else if (key.equals(IS_A) ||
>>                                        key.equals(RELATIONSHIP) ||
>>                                        key.equals(DISJOINT_FROM) ||
>>                                        key.equals(INTERSECTION_OF) ||
>>                                        key.equals(SUBSET)) {
>>                                //TODO: deal with relationships
>>
>>
>>                        } else if (key.equals(COMMENT)){
>>
>>
>> Peter
>>
>> On Jul 23, 2009, at 7:15, JP wrote:
>>
>>> Hi there at Biojava,
>>>
>>> I have an ontology file (from www.geneontology.org,
>>> gene_ontology.1_2.obo).
>>> A typical entry for a term is:
>>>
>>> [Term]
>>> id: GO:0000025
>>> name: maltose catabolic process
>>> namespace: biological_process
>>> def: "The chemical reactions and pathways resulting in the breakdown of
>>> the
>>> disaccharide maltose (4-O-alpha-D-glucopyranosyl-D-glucopyranose)."
>>> [GOC:jl,
>>> ISBN:0198506732 "Oxford Dictionary of Biochemistry and Molecular
>>> Biology"]
>>> subset: gosubset_prok
>>> synonym: "malt sugar catabolic process" EXACT []
>>> synonym: "malt sugar catabolism" EXACT []
>>> synonym: "maltose breakdown" EXACT []
>>> synonym: "maltose degradation" EXACT []
>>> synonym: "maltose hydrolysis" NARROW []
>>> xref: MetaCyc:MALTOSECAT-PWY
>>> is_a: GO:0000023 ! maltose metabolic process
>>> is_a: GO:0046352 ! disaccharide catabolic process
>>>
>>> I am reading this with the code suggested in:
>>> http://biojava.open-bio.org/wiki/BioJava:CookBook:OBO:parse
>>> I would like to get the is_a entries (as Term) - is this possible ? I
>>> tried
>>> to find this everywhere (annotations?) but find it (google searches
>>> included).
>>>
>>> Many Thanks
>>> JP
>>> _______________________________________________
>>> Biojava-l mailing list  -  Biojava-l at lists.open-bio.org
>>> http://lists.open-bio.org/mailman/listinfo/biojava-l
>>
>> Peter E. Midford
>> Mesquite Developer
>> Peter.Midford at gmail.com
>>
>>
>>
>>
>>
>
> Peter E. Midford
> Mesquite Developer
> Peter.Midford at gmail.com
>
>
>
>
> _______________________________________________
> Biojava-l mailing list  -  Biojava-l at lists.open-bio.org
> http://lists.open-bio.org/mailman/listinfo/biojava-l
>