[Bioperl-l] GeneStructure & Promoter(s)

Osborne, Brian Brian.Osborne@osip.com
Mon, 19 Feb 2001 09:31:28 -0500


> > related question is then also what distinguishes promoters() from
> > regulatory_elements().

Ask 10 molecular biologists a question, get 10 different answers. But on
this point I'd guess 7 out of 10 biologists would agree that a promoter is
that region of the gene 5' to the start codon that's necessary for
transcriptional initiation. This region usually contains the TATA consensus
(TATA box). It is typically separated from regulatory elements, which are
usually more 5'. There will be exceptions. A regulatory element may be
positive (e.g. enhancer, UAS) or negative (e.g. URS, repressor binding

These distinctions are not arbitrary, but based on the precise dissection of
the 5' region. For example, one might remove a UAS and render the gene
uninducible, but still transcribed. But if one removes the promoter the gene
can't be transcribed, irrespective of the UAS. These sorts of studies
indicated that the promoter is a binding site for RNA polymerase and
associated initiation factors, and positive regulatory elements are not.
These inferences were later proven by in vitro biochemistry.

There are situations where a gene has multiple start sites of transcription,
and these can be explained by multiple promoters or TATA boxes, in mammalian
cells in particular (in yeast there can be multiple start sites and a single
promoter). These start sites could be used in different biological
conditions, thus they are regulated, giving rise to one the exceptions I
mentioned. The regulatory elements and the promoter may be so interdigitated
in these situations that they cannot be mutated independently - it may be
difficult to tell whether they overlap or are the same.
There was also a question about regulatory elements and transcripts. There
are a number of well known cases of such elements in transcribed regions.
One of the most famous, perhaps one of the first, is the immunoglobulin
"heavy chain*" enhancer in its first intron. So this DNA element is a
classic transcriptional enhancer.

Regulatory regions have also been found in the transcribed 3' ends of genes,
and the problem for you here is that this regulation may be transcriptional
or post-transcriptional. For example, the iron regulatory element (IRE) of
the mammalian ferritin mRNA regulates the stability of the mRNA - more iron
and the mRNA is more stable, and this depends on the regulatory sequence in
the mRNA. Biologists consider these RNA elements to be true regulatory
elements, though they are not that common.

I apologize if I've been too didactic.

Brian O.

* The immunoglobulin is a tetramer, two larger identical "heavy" peptides,
two smaller peptides

 -----Original Message-----
From: 	Hilmar Lapp [mailto:hilmarl@yahoo.com] 
Sent:	Monday, February 19, 2001 3:46 AM
To:	Ewan Birney
Cc:	Bioperl; Ensembl Developm.
Subject:	Re: [Bioperl-l] GeneStructure & Promoter(s)

Ewan Birney wrote:
> On Sun, 18 Feb 2001, Hilmar Lapp wrote:
> >
> > In practice this means the following:
> > 1) We have to decide now whether to return an array or a scalar in
> > promoters(). My point was that Ensembl re-implements anyway, and
> > can enforce in the implementation that only a single promoter
> > feature be added. Is this okay with everyone?
> Who *wants* more than one promoter per transcript? 

A gene prediction result with suboptimal predictions included
(which BTW would also include overlapping exons; purely
biologically you would also disallow this, wouldn't you). Sorry,
that's my use case in this respect, and I want the model to cover

> > 3) We better decide now whether to add a property
> > regulatory_elements(), returning e.g. a GeneRegulatorI object. BTW
> > this could then also be the object returned by promoters(). A
> > related question is then also what distinguishes promoters() from
> > regulatory_elements().
> >
> This is not such a bad idea.

Thanks Ewan :-)) I've waited for this for months, you saved my day

> I guess regulatory_elements in my view would
> be on the gene (not the Transcript). But this is serious "no one really
> knows what is going on" terroritory. It would not kill me if we did not do
> this at all...

Quite encouraging. I'll wait and hear what other Ensembl folks
like to think before I dump the idea. (BTW I agree for gene
instead of transcript.)


Hilmar Lapp                              email: hlapp@gmx.net
GNF, San Diego, Ca. 92122                phone: +1 858 812 1757
Bioperl-l mailing list