[Go] addition of localization specific process terms ?

Wed Mar 4 01:56:27 PST 2009

Yes - I don't have much to add, but I agree with everything Val says 
(including what she attributes to me ;) ).

m

On Wed, 4 Mar 2009, Valerie Wood wrote:

> Because of all of the arguments in favour  mentioned by Karen and Chris I 
> thought it was always necessary and required for curators to make the more 
> granular annotation in these cases. We decided long ago that proliferation of 
> the ontology was not an issue when pitched against accurate capture of 
> biology, and  I wasn't aware that it was ever GO philosophy not to capture 
> compartment specific processes in this way.
>
> Another 'for' arguement is that, the more specific terms can have additional 
> parentage. For example, mitochondrial DNA replication, can have parents to 
> mitochondrial genome maintenance and mitochondrial organization, which would 
> not be possible if there was no mitochondrial replication term. This would 
> make lots more work as you would need to make concurrent annotations to 
> 'mitochondrial genome maintenance' and 'DNA replication". 
> I spoke to Midori about after the meeting and apparently  the terms which it 
> was agreed to purge from GO for this reason (very early in GO) were those 
> which combined Component and Function (not Process), so this may be the 
> source of the confusion. Apparently some  mitochondrial processes got purged 
> at the same time (i.e mitochondrial translation), but these were subsequently 
> added back.
>
> Val
>
>
>
> Chris Mungall wrote:
>
>> 
>> First, just to emphasize an important point:
>> 
>> An annotation to a term "P that occurs_in C" carries *more*  information 
>> than two independent annotations to P and C. This can be  seen if we have 4 
>> annotations, to P1, P2, C1 and C2. Currently we have  no way of knowing if 
>> this means that P1 occurs_in C1 or C2 or neither  or both, and similarly 
>> for P2.
>> 
>> Thus we need a way to annotate to "P that occurs_in C" if we are to  begin 
>> to fully capture the biology. There are two options
>> 
>> 1. post-compose
>>     http://wiki.geneontology.org/index.php/Annotation_Cross_Products
>>     - annotate to P and C as normal, using two annotations
>>     - in the annotation for P, put "occurs_in(C)" in col17
>> 
>> 2. pre-compose: create a term "P that occurs_in C"
>>     (and optionally for now, but required in the future, add the XP 
>> definition[*])
>> 
>> So your question is: how do we know when to choose 1 vs 2?
>> 
>> My own personal guideline is whether or not the location of P is an 
>> 'accidental' property of P, or something fundemental.
>> 
>> For example: translation in the forebrain is (I am guessing) not 
>> fundamentally different from translation in the midbrain. The cellular 
>> machinery goes through roughly the same sequence of steps using the  same 
>> materials. You can imagine 'transplanting' the process. There's  not much 
>> you would say in the definition of this term other than -  translation that 
>> occurs in the forebrain.
>> 
>> However, mitochondrial translation is fundamentally different from  nuclear 
>> translation because a different genetic code is used.
>> 
>> This argues strongly for "forebrain translation" being annotated by 
>> composing the description at annotation time (annotation xp), and 
>> mitochondrial translation being pre-composed in the ontology (and 
>> logically defined in bp_xp_cc).
>> 
>> Of course, I deliberately chose a clear cut example. Other cases will  be 
>> more difficult. But this isn't a worry, because the two methods are 
>> logically equivalent, so long as we keep bp_xp_cc up to date, and are 
>> clear about which relations to use.
>> 
>> Cheers
>> Chris
>> 
>> [*] see 
>> http://wiki.geneontology.org/index.php/XP:biological_process_xp_cellular_component 
>> 
>> On Mar 3, 2009, at 3:16 PM, Karen Christie wrote:
>> 
>>> Hi,
>>> 
>>> In last week's Reference Genomes annotation jamboree, a question came
>>> up about whether/when it is appropriate to add localization specific
>>> process terms. We'd like clarification of the GOC's current practice
>>> on this issue. More details are below.
>>> 
>>> thanks,
>>> 
>>> -Karen
>>> 
>>> 
>>> 
>>> We were discussing the appropriate process terms for the LONP1 group
>>> of genes, which are involved in proteolysis of proteins in the
>>> mitochondrial matrix. The question came up regarding whether it would
>>> be appropriate to have a specific term for mitochondrial proteolysis,
>>> considering that we already have terms like:
>>> 
>>> - ER-associated protein catabolic process (GO:0030433)
>>> - mitochondrial translation (GO:0032543)
>>> 
>>> At one time, the philosophy on this was to have one term representing
>>> the basic process, and to indicate localizations using the component
>>> annotations. However, more recently, terms such as the two listed
>>> above have been added to GO.
>>> 
>>> 
>>> Arguments in favor of localization specific process terms included:
>>> 
>>> -- 1. the fact that a different set of genes is involved in a given
>>> "process", e.g. mitochondrial translation versus cytoplasmic
>>> translation
>>> 
>>> -- 2. being able to look for terms enriched in a process such as
>>> "mitochondrial translation", without having to also use a component
>>> term to narrow the search to the mitochondrial genes (this is
>>> apparently not possible in many tools)
>>> 
>>> 
>>> Arguments against having localization specific process terms included:
>>> 
>>> -- 1. unnecessary proliferation of terms in the ontology, since the
>>> localization can be obtained via the component annotations
>>> 
>>> 
>>> So, for going forwards, we'd like clarification of when it is or is
>>> not appropriate to request localization specific process terms and on
>>> the guiding principles for deciding when/if such terms are  appropriate.
>>> 
>>> thanks,
>>> 
>>> -Karen
>>> 
>>> _______________________________________________
>>> Go mailing list
>>> Go at geneontology.org
>>> http://fafner.stanford.edu/mailman/listinfo/go
>>> 
>> 
>> _______________________________________________
>> Go mailing list
>> Go at geneontology.org
>> http://fafner.stanford.edu/mailman/listinfo/go
>> 
>> 
>> 
>
>
>
>