[Bioperl-l] Bio::Cluster

[Hilmar Lapp]

Furthermore, Bio::ClusterI has a concept of members, but promises no type. Bio::Cluster::SequenceFamily implements it by more or assuming it's going to be Bio::SeqI objects, but there's no checking nor enforcement.

Should there be a Bio::Cluster::MemberI? Or alternatively in the spirit of AnnotationCollectionI, should be have implicit type setting and enforcement in that the first added member determines the type members need to at least comply with, unless you explicitly specify the type that is at least required?

	-hilmar

> -----Original Message-----
> From: Hilmar Lapp 
> Sent: Thursday, October 24, 2002 3:16 PM
> To: bioperl-l@bioperl.org
> Subject: FW: [Bioperl-l] Bio::Cluster
> 
> 
> I'm more or less through this.
> 
> I also added a method display_id() in Bio::ClusterI. 
> unigene_id() and family_id() in Cluster::UniGene and 
> Cluster::SequenceFamily, resp., alias to this.
> 
> My feeling is many cluster implementions will want to be or 
> are naturally IdentifiableI. Would anyone have a problem with 
> me implementing IdentifiableI and DescribableI in 
> Cluster::UniGene. I'm not so sure regarding 
> Cluster::SequenceFamily, but my gut feeling says 'you too.'
> 
> I'm also going to implement Bio::ClusterI in 
> Bio::Cluster::UniGene (it wasn't so far).
> 
> Last but not least I'd propose to add annotation() to 
> Bio::ClusterI. Things like gene() etc in Cluster::UniGene 
> could then come off the annotation collection. Alternatively 
> I could write an adaptor like I did for Seqfeature tag/value 
> pairs that would give me a AnnotationCollectionI view on 
> Cluster::UniGene.
> 
> Comments/suggestions more than welcome.
> 
> 	-hilmar
> 
> -----Original Message-----
> From: Hilmar Lapp 
> Sent: Thursday, October 24, 2002 12:20 PM
> To: bioperl-l@bioperl.org
> Subject: [Bioperl-l] Bio::Cluster
> 
> 
> The modules and interfaces in Bio::ClusterI and 
> Bio::Cluster::* deviate in 
> a couple aspects from the rest of bioperl (or at least from 
> how the rest 
> of bioperl should be :) , namely
> 
> - interfaces traditionally do not have any implementation (I 
> know there 
> are exceptions, but there are usually good - or not so good - 
> reasons for 
> it)
> - as a consequence, there is no new() in interfaces
> - array accessors or iterators don't exhaust their arrays
> 
> I'm going to fix those things as I come across them, unless 
> anyone has 
> objections.
> 
> 	-hilmar
> -- 
> -------------------------------------------------------------
> Hilmar Lapp                            email: lapp@gnf.org
> GNF, San Diego, Ca. 92121              phone: +1-858-812-1757
> -------------------------------------------------------------
> 
> _______________________________________________
> Bioperl-l mailing list
> Bioperl-l@bioperl.org
> http://bioperl.org/mailman/listinfo/bioperl-l
> _______________________________________________
> Bioperl-l mailing list
> Bioperl-l@bioperl.org
> http://bioperl.org/mailman/listinfo/bioperl-l
>