Re: [Obo-discuss] Best practices?

[Chris M. <cj...@fr...>]

On Nov 21, 2006, at 3:30 AM, Richard Cote wrote:

> Hello all
>
> I just want to raise a discussion point about two ontologies  
> currently in
> the OBO foundry: attribute_and_value.obo and quality.obo. I had a  
> cursory
> look at the files and I'm wondering if what I'm seeing is what is  
> actually
> intended and if so, does it break ontology best practices?

Hi Richard

I think the problems stems from you using attribute_and_value.obo,  
which is deprecated - you'll see the obo metadata and front page  
refers to quality.obo

> The questions I have are:
>
> - can one term be in more than one namespace? (for example, PATO: 
> 0000001 is
> in namespace attribute and also in quality) What's the significance  
> of this?
> If you look at GO, which has 3 namespaces, I believe that one term  
> is only
> assigned to one namespace (though I'm not 100% sure about this).

Yes, each term belongs to a single namespace (ontology) in GO

> - if a term is defined with a relationship in one file, should it  
> not have
> the same relationships in the other? See PATO:0000004 in the text  
> below.
> Though I can see the arguments that term X might logically and  
> semantically
> have different relations in namesapce Y vs namespace Z, in  
> practice, it just
> makes everything confusing and hard to handle on a day-to-day basis  
> as most
> applications that use the ontology will not keep track of the  
> namespace (or
> to be honest, even care about it)

Again, I think you have a copy of an old file

In actual fact, an obo file can augment another obo file with  
additional relationships (but it cannot retract them). This goes for  
owl too.

However, it is good practice for an ontology maintainer to make  
available the entire ontology in a single file. Good practice has  
been followed in this case, it's just you have an old copy of a file

> - should one term id should have a constant name, or can it be  
> variable
> according to namespace? Case and point, look at PATO:0000002 in the  
> file
> exerpts below. Is this just a simple oversight, or is this  
> voluntary? If so,
> doesn't that open a big old can of worms for the reasons I've  
> mentioned
> above re: namespace usage in real life?

[snip]

again, ditch attribute_and_value.obo and you should have no problems.

Note the cvs directory may have old files littered around (we'll try  
and get round to removing them). But this shouldn't matter if you use  
the existing metadata file or the main web page.

The existing metadata file does have limitations - we haven't  
indicated that attribute_and_value has been superseded by  
quality.obo. In a few months we'll have a new system in place that  
should make some of this clearer.

Cheers
Chris

> Cheers,
> Rc
>
> --
> Richard Cote
> Software Engineer - PRIDE Project Team (Sequence Database Group)
>
> European Bioinformatics Institute
> Wellcome Trust Genome Campus            RC...@eb...
> Hinxton
> Cambridge CB10 1SD                      Phone: (+44) 1223 492610
> United Kingdom                          Fax  : (+44) 1223 494468
>
>
> ---------------------------------------------------------------------- 
> ---
> Take Surveys. Earn Cash. Influence the Future of IT
> Join SourceForge.net's Techsay panel and you'll get the chance to  
> share your
> opinions on IT & business topics through brief surveys - and earn cash
> http://www.techsay.com/default.php? 
> page=join.php&p=sourceforge&CID=DEVDEV
> _______________________________________________
> Obo-discuss mailing list
> Obo...@li...
> https://lists.sourceforge.net/lists/listinfo/obo-discuss
>