[MOBY-l] Queries in bioMOBY

[Chris Mungall]

On Fri, 8 Nov 2002, Lukas Mueller wrote:

> The fun thing with their implementation is that they seem to treat
> algorithms the same they treat simple services such as sequence
> retrieval etc. So ncbi blast is treated like another sql table (as ncbi
> sequence retrieval would be). I thought that was really cool (if I
> really understood it correctly!-).
> Cheers
> Lukas

let's see, how would that work - let's say you have a table (easier to
think of it as a relation/predicate) for getting blastp hits (ignoring
hsps for now), like this:

blastp(qseq, db, arguments, score, sseq).

you would have to force some arguments to be bound - the db, arguments,
and one of the sequences; just "select * from blastp" couldn't be done
sensibly in this universe.

SELECT * FROM blastp
WHERE score > 100
AND db = 'nr'
AND arguments = '-filter SEG+XNU'
AND qseq = 'MVVQQ..............';

I like this idea a lot! Makes a lot more sense if you think in
prolog/datalog terms.