[Bioperl-l] parsing tblastn using BPlite

Jason Eric Stajich jason@cgt.mc.duke.edu
Fri, 26 Oct 2001 23:03:24 -0400 (EDT)


Steve C wrote:

> Check out this post for some of the issues that I'm pondering:
> http://bioperl.org/pipermail/bioperl-l/2001-August/006131.html
>

Doh, remembered this post but didn't get a chance to follow back up with
you - figured my shot across the bow talking about starting to do a
SearchIO system would get reponses in time before I started in earnest,
but I really wanted to start work on the blastxml parsing if possible.

> A nice feature of the Bio::Search system is that there's a natural framework
> for segregating the different aspects common to all search-type algorithms
> (Processor, Result, Hit, and though it's not there yet, Run).
>
My thoughts exactly - I am just decoupling Search Parsing from Search
Result objects - Bio::SearchIO is event based parsing with a
Bio::SearchIO::SearchResultEventBuilder object handling these events and
building the correct Bio::Search objects.

> Perhaps we should both check in what we have so we can coordinate our
> efforts better. (I could also use another pair of eye/hands; my progress of
> late has been rather choppy as I find myself spending most of my free time
> chasing after an energetic 9-month old. 8)
> Steve

Sorry for not coordinating better with you - had a need - decided to go
ahead and write the code for it first.  Wasn't really sure where you were
and I assumed that more important life issues were dominating your time.
The object work you have done may be a more thorough treatment of the
necessary pieces for result parsing, I've no objections to chucking my
current objects if you have a better proposal for any of these (including
names) - I currently have:

Interfaces:
Bio::Search::ReportI
Bio::Search::SubjectI
Bio::Search::HSPI

implementations
Bio::Search::Report
Bio::Search::Subject
Bio::Search::HSP

Perhaps these should be better named and aggregated into the classes that
are there - oops.

I guess the best thing is to either post your code or commit it with the
caveat that it is under development and may not work (hence the 'live'
notion of the head of the dev tree).  Your call as to how to best start to
integrate our work, I'm more than happy to provide an additional set of
eyeballs to your code.

I have more things in the pipeline: Basic set of objects for working with
Maps and Markers, Phylogenetic Trees, and eventually Populations if anyone
has interest in working together on these feel free to post to the list.

-jason
-- 
Jason Stajich
Duke University
jason@cgt.mc.duke.edu