[Bioperl-l] Proposal: SemanticMapping and call for info on GeneObjects

David Block dblock@gnf.org
Wed, 15 May 2002 10:29:24 -0700


> -----Original Message-----
> From: Matthew Pocock [mailto:matthew_pocock@yahoo.co.uk]
> Sent: Wednesday, May 15, 2002 10:15 AM
> To: Hilmar Lapp
> Cc: Chris Mungall; Jason Stajich; dblock@gnf.org; Bioperl
> Subject: Re: [Bioperl-l] Proposal: SemanticMapping and call 
> for info on
> GeneObjects
> 
> 
> Hilmar Lapp wrote:
> 
> > That's a perl implementation right? :)
> > 
> > BTW has anyone checked where BioJava stands in this regard? 
> How does actually Apollo deal with that, i.e., does Apollo 
> have a class hierarchy representing Genes and constituents? I 
> think Dave is checking into that. Dave?
> > 
> 
> 
> > 	-hilmar
> 
> Hum.

Thanks for jumping in, Matt.
> 
> We have the interfaces for things like genes (in 
> org.biojava.bio.seq.genomic). We have the flexible pipe-lined 
> sequence 
> IO API (SeqIOListener & SeqIOBuilder & SeqIOFilter). We haven't yet 
> written the sequence IO handlers that intercept flat feature creation 
> events (like CDS) and emit tree feature creation events (like gene 
> models). 

Does Apollo use any of this code?

We would extends SeqIOFilter and it would consume feature 
> creation requests of particular types untill the endSequence 
> method was 
> called. Then, it would pattern-match its internal data-structure of 
> requests against some rules, pass on the feature creation requests to 
> the wrapped listener, clear any caches and then lastly pass on the 
> endSequence notification. The thing it passes events on to may be 
> another feature re-writing filter, or something else, so we 
> could layer 
> rules and transforms arbitrarily deep. I think that's similar to the 
> bioperl plan.
> 
> This all sort of smells like xslt.
> 

Absolutely it does, but we don't want to do it with xml, we want to do it in-memory.  Now, if we were to write a parser that threw SAX events based on Seq objects, we could reuse XSLT - right?  There's a job for a perl-xml junkie.

And the benefit of that would be that we could use the same xslt across languages, right?

<back to reading Java & XML/>

Dave

> Matthew
> 
>