[Bioperl-l] Re: [Bioperl-announce-l] an extension to Bio::SeqIO

Chris Mungall cjm at fruitfly.org
Wed Jun 18 10:19:41 EDT 2003



On Wed, 18 Jun 2003, Peili Zhang wrote:

>
> >> >>
> >> >> sequence data in any (rich) formats
> >> >> 	 |
> >> >> 	 | via Bio::SeqIO
> >> >> 	 v
> >> >     Bio::Seq->get_SeqFeatures()
> >> >         OR
> >> >      Collection of Bio::SeqFeatureI
> >> >
> >> >[Actually what's cooler I think is that you don't need Bio::Seq objects or
> >> >anything, just a set of Bio::SeqFeatureI objects. This would mean that
> >> >people could take their GFF files and turn them into chado IFF they are
> >> >rich enough.]
> >> >
> >>
> >> we do want the Bio::Seq objects. for instance, if the Bio::Seq object is a
> gene,
> >> as loading in its references as feature_pubs, accessions as feature_dbxrefs,
> >> comments/descriptions/others as featureprops. its gene model features
> >> (transcripts, exons, CDS's) will hang off of the top-level feature.
> >
> >
> >Okay but the part that builds the gene doesn't need all of this, it just
> >builds genes from a collection of features in the sort of multi-step
> >process that Chris described.  One would then transfer the pertinent
> >annotations,references, etc from a Bio::Seq object to a GeneStructure
> >object, but you don't need them to build the Gene object.
> >
> you're right that we don't need the annotation stuff to build the Gene object.
> but we don't want to lose any data during the data load. and I was hoping the
> GeneStructure object will keep the annotations since it inherits
> Bio::SeqFeature::Generic which in turn inherits Bio::Annotation::Collection etc.
> is this right?

I think we want to leave the Bio::SeqFeature::Gene::* classes out of the
{genbank,gff,whatever}->chado path

first of all, the core SeqFeature objects with typing and recursive
nesting of features should be rich enough for what we need.

Bio::SeqFeature::Gene::* doesn't seem well supported; I'd like to help
here but I don't know what the timeline would be. There could be some
flux, and there will probably be a lack of synchrony with SO for a while.

> -peili
>
>



More information about the Bioperl-l mailing list