[Go] query on annotation of protein binding and its children

[Harold Drabkin]

If you know that it IS a transcription factor (there is experimental 
evidence (not necessarily that paper), I would do it. And David adds 
that if  the target is a TF,  and it currenlty  doesn't have an 
experimental GO  annotation of some sort to that effect, we go out and 
make one. Otherwise, we would leave it GO:0005515 since there is no 
experimental proof that it is a TF.

As an aside, I find the children to be of great use when the target 
cannot be specified by a specific id.
Such as "actin_binding activity" if the paper doesn't say WHICH actin. 
In this case,  I use IDA
hjd


Rebecca Foulger wrote:
> Hi,
>
> Getting back to my GO, I've got an action item to move one of the unresolved queries from the last GO annotation jamboree onto email; so here goes...
>
> If protein x binds protein y, we want to annotate both proteins with protein binding; GO:0005515, or ideally one of its children (eg transcription factor binding/chaperone binding etc etc).
>
> To assign protein x with 'transcription factor binding ; GO:0008134', for example, how much evidence do we need that protein y is a transcription factor? Is sequence similarity to other transcription factors sufficient, or does it need to have been shown experimentally to be a transcription factor?
>
> This came up briefly on the conference call. The main issues were 
>
> 1/ how useful is it to use one of the children terms instead of GO:0005515.
>
> 2/ whether the children terms should only be used if protein x was involved in the related process (ie involved in transcription to annotate to transcription factor binding, or involved in protein folding to annotate to chaperone binding). The issue with this is obviously that alot of the time from straight protein-binding assays, you just don't know.
>
> 3/ Do you annotate protien x to E.g chaperone binding, when protein x is the target for the chaperone (as opposed to helping the chaperone fold a third protein)?
>
> Any thoughts/answers/queries to get some guidelines on this one?
>
> thanks,
> Becky
>
>   
> ------------------------------------------------------------------------
>
> _______________________________________________
> Go mailing list
> Go at geneontology.org
> http://fafner.stanford.edu/mailman/listinfo/go
>