Re: SenseLab Ontology

Hi Alan,

Thanks for your comments, many of the things you mentioned are already on my to-do list. Of course, some things (like drastic changes in nomenclature) have to be coordinated with the core SenseLab working group. Furthermore, some of the constructs are the result of an automated export. We don't want to make transient changes to the ontology that will be lost when the next automated export is made -- we will make these changes, but need some more time to implement them.

It seems the version that was online until now was not the very latest version of NeuronDB OWL. I have corrected that, and some of the things you pointed out are already corrected in the version that is now online.

>     - Similarly Transmittors. Ions > CO

CO is commonly seen as a gaseous transmitter, not an ion.


>   - Cell types could be linked to/ replaced with CL terms. If the  
> terms are not yet in CL, we have an offer from the CL curators to add  
> them. I'll send them the list.

With CL you mean the OBO Celltype ontology?


>   - Some of the Brain Regions, don't seem to be, e.g. retina,

This is a bit controversial. Many people see the retina and the optical tract as parts of the brain because of their ontogenesis. On the other hand, braininfo/neuronames makes the clear statement that the retina is not part of the brain.

> 
>    - Neuron can be below CARO_0000013 (cell)
> 
>    - Cell compartment can be below CARO_0000014 (cell component)

I will include this into the mapping file for now. Maybe we can integrate some selected classes of CARO into the SenseLab OWL file (without importing the whole ontology). Needs discussion.

 
>     - Don't believe you want or need the class "neuron  
> property" (neuron receptor is a neuron property. not)

This is also on my todo list, but again, it requires discussion. I would really prefer to see this as a parthood relation, but there might be arguments for viewing this as a 'property' or a 'quality' of a neuron. Of course, in this case it would be problematic to represent the 'receptors' as physical structures.

 
>     - I think it would be better  to use OBO RO for the relations,  
> rather than the unofficial ifomis version.

I hope there will be an agreement on a single namespace soon. For example, Birnlex uses the ifomis namespace, while John's OWL representation of the Allan Brain Atlas uses the OBO namespace.
As the meaning of the properties are not confined to the biological domain in any way, I would prefer a namespace that does not have biological connotations (for psychological reasons). Therefore, I prefer the ifomis namespace at the moment. I guess it would be best for the Relation ontology (and probably also the BFO ontology) to use a PURL namespace that is as neutral as possible. However, this should probably be discussed on other mailing lists.

cheers,
Matthias Samwald







.
-- 
"Feel free" - 10 GB Mailbox, 100 FreeSMS/Monat ...
Jetzt GMX TopMail testen: http://www.gmx.net/de/go/topmail

Received on Friday, 6 April 2007 16:01:25 UTC