[Bioperl-l] ribosome binding sites (RBS)

[Fernan Aguero]

+----[ Genevieve DeClerck <gad14 at cornell.edu> (26.May.2005 19:16):
|
| Hi,
| 
| I'm looking for the right bioperl class to use to describe ribosome 
| binding site (RBS) features in a bacterial genome.
| 
| The closest thing I could find is Bio::SeqFeature::Gene::UTR, but it 
| seems that this is too broad for my needs because, as I understand, an 
| untranslated region (a 5' UTR) can contain more than one type of 
| untranslated feature. Unless I'm missing something, it looks like I'll 
| need to create an "RBS" class. I'm thinking that it will need to inherit 
| from Bio::SeqFeature::Generic. Inheriting from 
| Bio::SeqFeature::Gene::UTR wouldn't make sense because a UTR *has* RBS's 
| - RBS's are not UTRs.

Genevieve,

[ notice: I have not done this RBS mapping myself ] 

If you only need to map this feature, then
Bio::SeqFeature::Generic should be OK, you'll have all your
features, whatever they're called mapped onto specific
locations  on the genome.

If you want to have a structure in bioperl that would allow
you to map several features at once, and you want this
structure to represent (in a hierarchy) the underlying
biology, then Bio::SeqFeature::Gene seems the way to go
(more experienced bioperlers, please correct me if I'm
wrong). Though in this case, yes, it seems like there is no
specific RBS module, but perhaps you can use any other
module that is designed to map a 'signal' onto a sequence.

Looking at the GenBank/EMBL feature table definition, it
seems like Bio::SeqFeature::Gene is trying to map the
features relevant for a 'precursor_RNA', and these features
are not supposed to get nested (they don't overlap).

http://www.ncbi.nlm.nih.gov/collab/FT/#7.3

The feature table def has RBS under 'misc_signal' together
with other 'signals' (ie features that logically have to be
nested within other feature). One possibility is to have RBS
be a subfeature of UTR. But I don't know how deep a
feature hierarchy can be in Bioperl and whether this can
then be mapped back easily to genbank, nor if you would need
to do that. 

Clearly there is no 1 to 1 mapping between bioperl modules
and genbank features (no Bio::SeqFeature::MiscSignal), but
perhaps there are reasons for this. Perhaps you can work
around this by reusing existing objects?
 
| Has anyone else described/instantiated RBS's with bioperl?
| 
| Any comments or insight would be appreciated.
| 
| Thanks,
| Genevieve
|
+----]

I'm sorry I can't be more specific. Hope this give you some
ideas.
 
Fernan