[Ontology-editors] [ geneontology-Curator requests-2053783 ] parentage: RNA modification

[Karen Christie]

Hi,

I wasn't suggesting that we shouldn't have the logical portion of the 
definition, just that I don't find them sufficient for the purpose of 
annotating. My concern was the idea that "Ideally we could appeal directly 
to the definitions without the need for subjective interpretation." Even 
when the logical defs are present and consistent with the tree, I'm not 
sure that it is ever going to be the case for annotating that subjective 
interpretation is not needed. When annotating, I often need more info than 
the logical def to translate what's in a paper to the appropriate 
corresponding GO term. However, it doesn't sound like anyone disagrees 
with the need for the 'gloss'. As an aside, the one time I heard Barry 
speak, he definitely sounded like his preference would be to completely 
remove the 'description/gloss' portion of the definitions.

As for agreement between GO, SO, other ontologies, I didn't mean to imply 
that I'm against that, just that I don't think GO should blindly follow 
another ontologies term name. In this case, I think that SO's term name 
choice seems rather odd, definitely not common usage in my reading, so 
perhaps SO is the one that should be changing, rather than GO, in this 
particular situation.

-Karen


On Thu, 21 Aug 2008, David Hill wrote:

> Hi Everyone,
>
> I must say that I agree with Midori. I think that the logical definition 
> makes it infinitely easier to develop the ontology and to check to make sure 
> that the terms are in the correct place and have the correct relationships. I 
> think the gloss is valuable when we are using the terms for annotation since 
> they often explain the differentia. This becomes very valuable when 
> annotating genes.
> David
>
> Midori Harris wrote:
>> Hi,
>> 
>> On Wed, 20 Aug 2008, Chris Mungall wrote:
>> 
>>> 
>>> Hi Karen
>>> 
>>> Hope you don't mind if I answer your points on the editors list as I think 
>>> they cut across the whole ontology. I would hope we can come to a common 
>>> agreement here
>>> 
>>> This relates to:
>>> https://sourceforge.net/tracker/?func=detail&atid=440764&aid=2053783&group_id=36855 
>>> 
>>>> 
>>>>> Ideally we could appeal directly to the definitions without the need
>>>>> for subjective interpretation.
>>>> 
>>>> To be honest, as an annotator, the part of the definition I usually
>>>> find most helpful is the part that Barry Smith dismissed as merely
>>>> descriptive and ideally not part of the definition. The 'logical' part
>>>> of the definition is rarely helpful to me.
>>> 
>>> I find it odd that you don't find this part helpful. I find the vast 
>>> majority of parentage issues in GO can be automatically determined by a 
>>> rigorous logical definition. I imagine the different parts of the 
>>> definition serve different roles for different purposes, and coming from 
>>> an annotator's perspective the descriptive part plays a different role 
>>> from say for example determining the correct parentage in GO. But surely 
>>> this is important to an annotator too?
>> 
>> Interesting ... my initial, intuitive reaction was not to be all that 
>> surprised by Karen's comment. Upon further reflection I wouldn't say that 
>> logical definitions are useless or uninformative to biologists, but that 
>> "necessary and sufficient" for logical purposes sometimes doesn't equate to 
>> "everything a biologist would want to know" about a term or the bit of 
>> reality it represents.
>> 
>>> I'm not sure Barry was dismissive of the "gloss" part of the definition. 
>>> Of course, it would easy to read the wrong sense of the term "gloss"
>> 
>> I wouldn't presume to read Barry's mind, but I do recall him saying things 
>> that could give the impression that he dismisses "gloss". But I don't think 
>> that's a major concern for us.
>> 
>> [snip 'gloss' defs]
>>> I like gloss. I would happy with massive wikipedia-article sized pieces of 
>>> gloss (structured such that the succinct summary is easily digested of 
>>> course).
>>> 
>>> I just think it's important to distinguish between the gloss and the 
>>> formal part of the definition, and to write the formal part of the 
>>> definition in a different style from the gloss. The formal part of the 
>>> definition should ideally be in genus differentia style and clearly and 
>>> succinctly state the necessary and sufficient conditions, using only 
>>> simpler terms, preferably using terms from the same ontology or other 
>>> ontologies. Of course, there are some terms for which this is difficult : 
>>> parts of the ontology on the cutting edge of science for example. But I 
>>> would bet that for the majority of terms it is achievable.
>>> 
>>> I don't have such strong opinions on the gloss part of the definition.
>> 
>> My 2-small-currency-units: Even with well-written formal definitions (which 
>> I agree we should have), for many terms a gloss will remain valuable -- for 
>> some terms, indispensable -- for annotators and other biologically-oriented 
>> users to understand what a term represents and how it should be used.
>> 
>>>>> The def of RNA processing introduces 2 terms that are not defined in
>>>>> GO: primary RNA transcript and mature RNA molecule. (not sure the
>>>>> switch between transcript and molecule). If we refer to SO here:
>>>>> 
>>>>> SO:0000185 !  primary_transcript [DEF: "A transcript that in its
>>>>> initial state requires modification to be functional."]
>>>>> 
>>>>> SO:0000233 !  mature_transcript [DEF: "A transcript which has
>>>>> undergone the necessary modifications, if any, for its function. In
>>>>> eukaryotes this includes, for example, processing of introns,
>>>>> cleavage, base modification, and modifications to the 5' and/or the 3'
>>>>> ends, other than addition of bases. In bacteria functional mRNAs are
>>>>> usually not modified."]
>>>> 
>>>> In my reading, I'm not sure if I've ever seen it called the 'mature
>>>> transcript'.  After maturation, it is generally referred to as a
>>>> 'mature RNA', e.g., the mature 25S rRNA or the mature 18S rRNA. My
>>>> sense is that the transcript is what is copied from the DNA and once
>>>> it's been fiddled with so that it isn't a direct copy, you wouldn't
>>>> refer to it as a transcript anymore.
>>> 
>>> I agree "mature transcript" sounds unusual, but I have heard "processed 
>>> transcript" a lot
>>> 
>>>> So, I'm really not sure that SO's
>>>> nomenclature is standard here, or that GO would want to follow it.
>>> 
>>> I think we want to use the same terms as much as possible, so I would hope 
>>> that GO wouldn't just ignore SO's terms and say "we don't want to follow 
>>> that". Instead we'd work together on the terms that are acceptable by all.
>> 
>> I agree -- it seems obvious that we should determine what name best 
>> balances accuracy and community usage, and use that name (plus any 
>> synonyms) in both GO and SO.
>> 
>>> This goes for all OBO ontologies of course. There is probably more we can 
>>> do to facilitate this kind of inter-ontology discussion, beyond mutual 
>>> cross-linked tracker items, ideas welcome!
>> 
>> Is there any centralized forum for this sort of issue? The obo-discuss list 
>> seems to broad in scope (it's not restricted to content (rightly), and 
>> tends towards higher-level issues), whereas many of the other obo-* lists 
>> are specific to an ontology or to topics that are cross-ontology but rather 
>> technical (e.g. obo-relations, which seldom fairs to make my head spin). 
>> I'm not even sure that Yet Another Mailing List is the way to go ...
>> 
>> m
>> 
>>> Off my soapbox, I may respond to some of the specific points below back on 
>>> the tracker...
>>> 
>>> Cheers
>>> Chris
>> _______________________________________________
>> Ontology-editors mailing list
>> Ontology-editors at geneontology.org
>> http://fafner.stanford.edu/mailman/listinfo/ontology-editors
> _______________________________________________
> Ontology-editors mailing list
> Ontology-editors at geneontology.org
> http://fafner.stanford.edu/mailman/listinfo/ontology-editors
>