[Biojava-l] status? Blast parser?

[Simon Brocklehurst]

> My personal bias for these parsers is to build a model like the XML sax/dom
> division of labour. So - a blast parser would fire events for hits or
> alignments & stuff, and simple apps can just listen for the items they want
> (blast version and top 10 hits for example), or you could create a blast
> document object from this event stream, and then manipulate the whole
> report. The event layer might not use any biojava concepts at all, where as
> the blast document would be composed from a SequenceDB object, sequence
> objects, locations, features & loads of other data-structures. Of course,
> you may have done all of this in a completely different way...

Our software is designed to work within our own framework which populates our
object models in the ways that we want.  Implementing SAX-like functionality
would be a really nice idea for people who might want to minimize memory useage
and use their own object models for some applications - and we could in
principle add a layer like this.

Before going down this route, however, we need to consider the point that
something like this already appears to exist.  See

http://workingobjects.com/blastxml/bioperl_blastxml.txt

I have no idea how good this software actually is, but if it works well,
mightn't this actually be the way to go for a biojava BLAST parser?  Is anyone
in contact with this guy? The link above is to a copy of a post to bioperl over
a year ago.

There seems no point in duplicating work if it's already be done before in a
high quality way.

What do people think?

Simon
--
Simon M. Brocklehurst, Ph.D.
Head of Bioinformatics
Cambridge Antibody Technology
The Science Park, Melbourn, Cambridgeshire, UK
http://www.CambridgeAntibody.com/
mailto:simon.brocklehurst@CambridgeAntibody.com