[Bioperl-l] Gene Interfaces in Bioperl 0.7; Review of bioperl 0.7 from Ensembl

Ewan Birney birney@ebi.ac.uk
Sat, 17 Feb 2001 14:19:50 +0000 (GMT)

Hilmar -

I like the interfaces. They are quite feature-rich but that does make
them useful. 

My only comment is that I would expect 


to return a single object. A transcript is, in my view, defined to have a
single promoter (ie, there is one unique starting point in the DNA). Of
course, this really depends what on earth is "a promoter" --- which we are
punting on right?

This model works very well with Ensembl. Once bioperl-0.7 is out and
stable we should be able to be compatabile  with it very quickly.

The only challenge here is implementing aspects of the SeqFeature
interface on our gene/transcript objects, which prefer to be "location
-less". This is ok in our new world of rich locations - it is just a
challenge to implement.

(one for Arne - it is like when we get Genes/Transcripts out from the
database currently they have *no context* in a location sense. Only their
exons have a context. The bioperl Gene/Transcript interfaces require a
gene to have a context, being start/end/strand/seqname so we have to dream
up what we can do here. Easiest fix is that context is the implied context
from the start-codon-exon.)

Re: Ensembl <-> Bioperl 0.7

I am pleased to say that with only one addition I was able to get Ensembl
to pass its entire test suite (390 tests) against Bioperl-live /0.7

I got alot of "deprecated object" noise etc etc...

I think that is a real tribute to Hilmar and Jason providing excellent
backward compatability for bioperl. We are indebted to you two guys.


Ewan Birney. Mobile: +44 (0)7970 151230, Work: +44 1223 494420