[Biojava-l] GFF & feature creation

[Ewan Birney]

On Tue, 22 Feb 2000, Peter Rice wrote:

> Matthew,
> 
> We are going through feature handling for EMBOSS too. Internally,
> we are keeping somethign similar to GFF but it raised some issues.
> 
> Proteins we will treat the same as DNA, but ignore the strand and
> frame fields in GFF.
> 
> Joins across sequences are a problem. For example, the following EMBL
> entry where all except one of the the exons (and flanking sequence)
> are in separate entries.
> 

Over here at bioperl/ensembl we too have this endlessly thorny problem.

I have come around to having a seqname attribute on sequence features
and *optionally* being attached to sequences. I am not sure Matt if
you have this system in biojava.


Matt's constructor problem is different and I appreciate that as well.
The many implementation/many type-interface problem is well... a
problem. I need to read up about this momento thing.


I would advise to try to keep as many things out of the sequence object
as possible though....



-----------------------------------------------------------------
Ewan Birney. Mobile: +44 (0)7970 151230
<birney@ebi.ac.uk>
-----------------------------------------------------------------