[Bioperl-l] Re: Bio::SeqFeature::OntologyTypedI Proposal

[Chris Mungall]

On Thu, 11 Aug 2005, Ewan Birney wrote:

>
>
> Hilmar Lapp wrote:
> >
> > On Aug 10, 2005, at 2:03 PM, Chris Mungall wrote:
> >
> >> Regarding 'magic' in SeqIO - not sure this is required. You can already
> >> plug in your own factories here, we just need to extend this with feature
> >> factories. The default method will continue to produce relatively light
> >> SF::Generics?
> >
> >
> > Right, this is exactly what I was thinking. A feature factory that
> > creates ontology-compliant features will also probably need to have
> > something like an OntologyTermResolver, in order to check a given
> > feature type (primary_tag) against an ontology that sits somewhere
> > (local file, local database, remote database, or even remote file).
> >
>
> Ok - I'll hold off the magic for the moment, but I think it would
> be nice to have just-enough of SO in-built into Bioperl so one
> could do something like:
>
>    $seqio = Bio::SeqIO->new( -file => "-", -format => 'EMBL', -feature_converter => 'SO');
>
> and the "right thing" happens.

actually, Bio::SeqFeature::Tools::TypeMapper already does this. Well, you
still have to wrap it to have the above work, but the mapping is there.

Of course, you can always provide your own mapping as a hash (which could
come from an ontology, a database, whatever). But like you say the gb->SO
type mapping is so common that it's good to have a default hardcoding
here.

> Does anyone want to propose an alt name to
>
>    Bio::SeqFeature::OntologyTypedI?
>
> But for that is it ok for me to implement and commit?
>
>
> >     -hilmar
>