[Biojava-l] executing blast nicely (was Blast)

[Jason Eric Stajich]

> Of course there are probably better ways to do this. And perhaps someone
> would have ideas about how to abstract it a bit so we could handle cross
> network activity as well as local system calls. If we can come up with a
> design, I would be happy to move my things over to biojava.
>

This is something I think all the language toolkit projects want to
address - through coordination of open-bio Ewan has suggested a sort of
/etc for bioinformatics so that we can devise a standard way to do this
type of thing.

For blast NCBI does this with a .ncbirc file listing paths to blast dbs,
I'd like us to think about developing the file which has the blastdbs,
blast exes, blast remote srvs, etc.

This is something I'd very much like for us to discuss at the ORA
conference if we haven't hashed it out enough over the list(s)  (perhaps
it should be hashed out by a team on a separate list which report back
some proposals to the larger groups).  We had an infrastructure list for
this sort of...

We do it fairly simply in bioperl, with a single of environment variables
$BLASTDIR which expects a blastall exe and a data dir with matricies and
blastdbs (fairly inflexible, but code that works).  Additionally I wrote a
module which will submit jobs through HTTP to NCBI blast srv
called RemoteBlast.  One could imagine another module that
wrapped a corba interface to openBSA job submission as well (something
we're tinkering with in biocorba but still trying to finish writing the
perl libs).

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