Re: blog: semantic dissonance in uniprot

On Mar 26, 2009, at 1:41 PM, Michel_Dumontier wrote:

> I found it with the EBI's OLS service
>
> http://www.ebi.ac.uk/ontology-lookup/
>
>
> oh, I was going to ask - what tool should we use for obo2owl nowadays?

Use the pre-generated OWL files linked from the obofoundry.org webpages

If you want to do a conversion yourself (i.e. on a locally modified  
obo file), you can use the REST-ish converter
http://www.berkeleybop.org/obo-conv.cgi

The results should be identical to the pre-generated ones

If you want you run this locally, you can download the xslt.

A few caveats

- this generates purl.org/obo URLs. At some point in the future these  
will become weakly deprecated and we will move towards  
purl.obofoundry.org URLs. There will be an interim period where we  
support both schemes at each URL, and equivalence axioms (possibly as  
a separate bridge file).

- XSLT doesn't scale well. If you want to convert ncbi_taxonomy.obo  
you'll need a lot of memory. At some point the XSLT will be abandoned  
in favour of the java OBO-API, which has a partially implemented  
bridge to the OWL API. If you want to be a bleeding edge user you can  
try this.

- The current mapping faithfully preserves all terminological metadata  
using the n-ary relations pattern. This introduces extra instances  
that look odd in some tools and may degrade the performance of some  
reasoners. The java converter gives you flexibility regarding whether  
you want to preserve or discard this metadata. When OWL2 is  
established the official mapping will switch to using axiom  
annotations instead of the n-ary relations pattern.

> -=Michel=-
>
>> -----Original Message-----
>> From: public-semweb-lifesci-request@w3.org [mailto:public-semweb-
>> lifesci-request@w3.org] On Behalf Of Chris Mungall
>> Sent: Thursday, March 26, 2009 4:35 PM
>> To: Michel_Dumontier
>> Cc: Kei Cheung; W3C HCLSIG hcls; Matthias Samwald
>> Subject: Re: blog: semantic dissonance in uniprot
>>
>>
>> On Mar 26, 2009, at 8:33 AM, Michel_Dumontier wrote:
>>
>>> The Gene Ontology says a Protein (GO:0003675) is a Thing (the entity
>>> was removed, but then later added so the identifier wouldn't be
>>> reused)
>>
>> More accurately: very early versions of the GO had a term named
>> "protein", but this was obsoleted because it was out of scope for the
>> GO. The term was obsoleted, but a historic record is maintained. GO
>> does not intend to say "protein is_a thing", what you are seeing is
>> probably an artefact of the translation of obsoletion metadata to  
>> OWL.
>
>
>

Received on Thursday, 26 March 2009 21:34:22 UTC