[go] 'regulation of gene expression'

[Harold Drabkin]

Jim Hu wrote:
> I'm relatively new to the group, but I think the term is useful if and 
> only if the target (direct or indirect) is specified.  From the pov 
> of  an experimentalist, I'd like to mine the annotations to find 
> associations that could be pushed down to lower nodes if more data was 
> available.  In other words, the ability of GO terms closer to the root 
> to identify gaps in the experimental literature is a good thing, since 
> it suggests places where more experiments are needed.
>
> A while ago I put up an SF query about whether terms that specify the 
> target in the term definition were too specific.  e.g.
> GO:0046011 ! regulation of oskar mRNA translation
>
> and it's two children would be lumped into regulation of mRNA 
> translation and given oskar mRNA as the target.  It sounds to me like 
> the discussion is coming around to my pov on this (I hope?!).  I can 
> see why with may not be appropriate, if the regulator is part of a 
> complex that does the regulation, then it seems like you could be 
> getting into having two kinds of with, the partner and the target.
>

This sort of came up many meetings ago, because we cannot put a new term 
into the GO for each and every specific *instance*.  (regulation of 
(your favorite) mRNA translation, or biosynthesis of ....fill in  your 
favorite item. Not only does this start down an infinite expansion of 
the size of the DAG, but also moves away from the idea of terms that 
would be useful to a broad range of MODs, not just one or two.
It isn't just the actual gene that you think is being regulated, it's 
whether your gene product is directly participating in that regulation 
process.

an aside:.....

We also discussed what and when to put something in the with field; the 
idea was to ask the question "is there something missing if you turned 
the annotation into a sentence... like, Product X has kinase activity 
based on sequence similarity.  The question would be "similarity to 
what?", wich is answered by the WITH field.

IPI, IMP, and IGI all beg a question .. to/with/of WHAT? (based of 
physical interaction with WHAT; based on (WHAT) mutant phenotype?; based 
on genetic interaction with WHAT.

I guess, if you use IEP, you could ask " .. based on the expression 
pattern of WHAT?"

(I realize the paradigm breaks with IEA and RCA, as the question asked 
would be more along the lines of WHAT KIND (of electronic annotation or 
computational analysis; by this one could make (not that I am) an 
arguement that IDA would invoke WHAT KIND (of direct assay). ..

but I digress..

The point is just how much can we put into an annotation?  How much can 
we distill an experiment into a single line of 15 tab-delimited fields? 
How much should we expect a user to refer to the PMID of the annotation 
if it is of interest to them?


> Adding a new field presumably opens another can of worms, but it 
> sounds like maybe one is needed.  But here I'm wading into waters best 
> left to more experienced GO people.
>
> Jim
>
> On Aug 10, 2007, at 9:42 AM, E Dimmer wrote:
>
>> Hi,
>>
>> There is a large number of terms in GO which deserve more information 
>> regarding the 'target' of its gene product's activity, for instance:
>> -- negative regulation of ubiquitination
>> -- protein kinase activity
>> -- regulation of DNA binding
>> -- regulation of phosphorylation
>> -for all of these examples I have information on the specific target 
>> of a protein's action.
>>
>> While I strongly feel that this target information should be captured 
>> -- is the 'with' the right place? We seem to have tried to be very 
>> strict on the contents of the 'with', so the type of value used is 
>> closely related to the type of evidence code applied in the 
>> annotation. I'm not sure that regulation of expression be an 
>> exception to this...
>>
>> Emily
>>
>>
>> Pankaj Jaiswal wrote:
>>>
>>>
>>> Harold Drabkin wrote:
>>>
>>>> Perhaps if used, one needs to restrict the evidence code used with 
>>>> it to IEP and perhaps also allow a With field dbx ref for a gene_id 
>>>> if known or suspected
>>>>
>>>
>>> With field filling should be 'a must' because the regulation of 
>>> expression is always in context to another gene (product). However 
>>> the code can also be IPI, IGI, IMP besides IEP.
>>>
>>> Pankaj
>>
>>
>> -- 
>> ************************************
>>    Emily Dimmer
>>    GOA Coordinator
>>    EMBL-EBI
>>    Wellcome Trust Genome Campus
>>    Hinxton
>>    Cambridge CB10 1SD, U.K.
>>    Tel:     +44 1223 494654
>>    Fax:    +44 1223 494468
>>    email:  edimmer at ebi.ac.uk
>> ************************************
>>
>
> =====================================
> Jim Hu
> Associate Professor
> Dept. of Biochemistry and Biophysics
> 2128 TAMU
> Texas A&M Univ.
> College Station, TX 77843-2128
> 979-862-4054
>
>