[Ontology-editors] question about cellular component organization (from bp-xp-cc evaluation)

[Midori Harris]

A bit of further thinking below ...

>> We've deliberately created separate terms for protein complex assembly and 
>> cellular protein complex assembly (with parallel structures for 
>> macromolecular complex assembly and m. c. subunit organization) because 
>> some complexes are assembled at the multicellular-organism level. The 
>> circulating lipoprotein complexes -- HDL, LDL, etc. -- are examples.
>> 
>> So it seems that, instead of asserting the implied link, we actually have 
>> to move cellular component organization and cellular component biogenesis 
>> directly under biological process, not under cellular process. This will 
>> fix a bunch of other implied links too, I think.
>> 
>> I've also put this on the wiki:
>> http://wiki.geneontology.org/index.php/XP:biological_process_xp_cellular_component#2009-06-16
>> 
>> Comments welcome!
>
> So the inferences here follow from definitions that use the has_level 
> relation. For example, cellular process = biological_process that has_level 
> cell.
>
> I'm really not sure the best way of defining has_level. But it seems that if 
> we use this relation in logical definitions for terms that say "cellular 
> <BP>" then the reasoner can at least make us consistent, as it has in the 
> case above.

What rang the alarm bells is that the reasoner was suggesting links to a 
"cellular" parent for terms that are not themselves "cellular <bp>", like 
the example I included:

> Assert: protein complex assembly is_a cellular macromolecular complex 
assembly

Obviously we ought to fix the inconsistency that's spawning those finds, 
rather than add the links as suggested. The 
biological_process_xp_cellular_component.obo file doesn't actually have 
any xps for "cellular component organization" or "cellular component 
biogenesis"; I think the problem is that both are is_a "cellular process".

But if we just move "cellular component organization" and "cellular 
component biogenesis" up to be direct is_a children of "biological 
process", some of the descendants will lose paths to "cellular process". 
The simplest solution is just to add the "cellular process" parent back to 
terms such as "cellular macromolecular complex subunit organization". A 
lot of fairly specific terms would end up directly under "cellular 
process" this way (e.g. "cytoplasm organization", "synaptogenesis"). Maybe 
we can live with that, although it would be neater to have "cellular <BP>" 
terms to group them. The catch with the better-organized approach is names 
for some of the terms -- "cellular cellular component assembly" sounds 
absolutely ridiculous, and words can't express how confused our users 
would be.

Any thoughts?

m