[Bioperl-l] Re: Promoter

[Samuel Aparicio]

Elia Stupka wrote:

> > hard). So I would suggest to keep these elements separate from the
> > promoter.
>
> I would argue the opposite, I would argue that we should be elastic even
> if perhpas inexact n biological sense (considering that the promoter field
> of science is prety inexact in itself).
>
> In other words, I think that promoter should be basically a SeqFeature
> container, so that at will people can implement SeqFeature compliant
> objects with more or less complexity, to describe their favourite promoter
> elements, and whatever complex behaviour they want to attrbiute to those
> elements. Whether the element is 10Kb upstream or not, I reckong it should
> be contained in the promoter object. In this way one can pull out from one
> object any information known about its regulatory elements.
>
> If this seems inapporpriate biologically we can argue at length about
> naming things,i.e. maybe it should be regulatory elements, or whatever, my
> point is that from the object point of view it would be nice to have
> everything in one object associated with the transcript.
>
> An obvious example of the usefuleness of this is if people want to start
> affing elements which are knwon to bind to each other and trigger
> something only if bound together. From a basic point of view, all you want
> is to ask what elements regulate this gene, and where are they? The other
> example being that you add some basic promoter element, then you discover
> a very strong enhancer 20Kb upstream, and just because we have a rigid
> object layer, we cannot simply  say there is also this enhancer
> upstream...
>
> hmm... hope that was clear enough....
>
> Elia
>
> _______________________________________________
> Bioperl-l mailing list
> Bioperl-l@bioperl.org
> http://bioperl.org/mailman/listinfo/bioperl-l

I can see your line of thinking from the object method point of view. However
I wonder
how this would handle the biological situation where an enhancer for one gene

is also the promoter for a second gene? (I have come across such a case in
practice).
Naming might be an issue - I might venture that you need to have a general
transcriptional
regulatory element class of which "promoter", "enhancer", "silencer",
"insulator", etc.
might be subsets.

Does the binding of regulatory element to transcript in the method then
*require* that every regulatory
element have a transcript? Again there are cases where arbitrary DNA
sequences are tested
for regulatory activity where their target of action is either complex or the
gene is unknown.

my $0.02

Sam



--
Dr Samuel Aparicio BMBCh PhD | saparici@hgmp.mrc.ac.uk
Wellcome Trust Centre for Molecular Mechanisms in Disease
Cambridge CB2 2XY. UK. | +44 1223 762663 (tel)