[Bioperl-l] GFF method in Bio::SeqFeatureI

Hilmar Lapp hilmarl@yahoo.com
Sat, 17 Feb 2001 15:02:11 -0800


Ewan Birney wrote:
> 
> On Fri, 16 Feb 2001, Hilmar Lapp wrote:
> 
> > There now two methods in SeqFeatureI for every gff-concerned
> > operation: gff_string() and gff2_string(), _from_gff_string(),
> > _from_gff2_string(), and two named parameters in new().
> >
> > Inheriting classes all use gff_string() and friend.
> >
> > I suggest we remove the old implementation of gff_string() and
> > rename Mark W.'s gff2_string() to the former. Likewise for the
> > other methods.
> >
> > Alternatively, we rename gff_string() to gff1_string(), and
> > gff2_string() to gff_string().
> >
> > I prefer the first option. Any objections?
> 
> I am good with outputting only gff2, but we need to accept gff1 - or
> indeed badly formatted gff in general. If I remeber rightly, Mark's
> _from_gff2_string is very "picky" whereas there is alot of "GFF" out there
> which is no where near the spec.
> 

I'm not sure but I think the historic gff_string() wouldn't parse
truely GFF2-compliant strings correctly. Is that right Mark?
(otherwise we wouldn't need _from_gff2_string())

So, your point leads to the question whether to have one single
_from_gff() that gracefully parses GFF1, bad GFF, and GFF2, or
have _from_gff2 and _from_gff1. The first option seems prone to
cause nightmares. Comments? Mark?

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