W3C home > Mailing lists > Public > semantic-web@w3.org > October 2006

RE: using OBO in owl format to describe data

From: Xiaoshu Wang <wangxiao@musc.edu>
Date: Wed, 11 Oct 2006 14:26:08 -0400
To: "'Mary Montoya'" <mhm@ncgr.org>
Cc: <semantic-web@w3.org>, <public-semweb-lifesci@w3.org>
Message-ID: <002b01c6ed62$b86bfaa0$4a741780@bioxiao>

-- Mary, 

> so I should define properties for my data so I can then state 
> ( pardon my awkwardness with the shorthand syntax ):

The notation is notation 3 (with a few variants).  Its terse syntax make it
much easier to communicate among humans. You can learn the language syntax
starting from http://www.w3.org/DesignIssues/Notation3. 

> @prefix gene: <http://example.com/genes#> .
> gene:Y rdfs:subClassOf obo:Term .
> _:x a gene:Y . 
>     rdfs:label    "my frog gene"
>     mystuff:sequence  "atgcgga...."
>     mystuff:chromosomeNumber   3
>     mystuff:startCoordinate    12345
>     mystuff:endCoordinate  7890
> so then if my friend at bullfrog genome project states
> @prefix gene: <http://example.com/genes#> .
> gene:Z rdfs:subClassOf obo:Term .
> _:w a gene:Z . 
>     otherstuff:sequenceID  "786451"
>     otherstuffchromoNum   "3"
>     otherstuff:comment    "source unknown"
>     otherstuff:type  "pseudogene"
> Both will be of type obo:Term but since obo:Term doesn't 
> define any properties for individuals what good does it do 
> me? I can't coordinate the data. 

First, the point of my last email is that there shouldn't be any obo:Term.
It is useless and should be eliminated. Second, even if you created such
term like obo:Term, which demands its instance to have an obo:name, obo:id,
etc., what if your friends do not provide them? You still couldn't
coordinate their data, right?  

Missing the obo:name or obo:Id does not make their genes an obo:Term, but it
still makes their gene a Gene.  Your purpose, however, is to interoperate
the knowledge about Gene but not the obo:Term, right?

If the data owner wants to share their data, they would use the most
commonly used vocabularies to describe it. If they don't, you cannot do
anything about it anyway. Using a pseudo-superclass gives you a false-sense
of ensured-interoperability. But, the true interoperability can only be
ensured by "sharing" ontologies.  

> I guess if I create some properties that I think will be 
> useful and assign their domain to be obo:Term and then tell 
> my friend to also use them to describe their instance data, 
> then our data will interoperate.  
> And you are saying that if I do a good job of identifying 
> useful properties over time many people will also start using them.

In a way, yes.  If you think you cannot find any existing ontology to serve
your need, create your own.  Publish it and try to pursuade others to use
it.  It does not have to be perfect, but you do need to make it popular.

Received on Wednesday, 11 October 2006 18:29:44 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 1 March 2016 07:41:53 UTC