keep .2bit reference in memory during batch blat jobs

[splaisan]

I succeeded to run a single blat search with standalone blat installed but would like to reduce the time of loading the reference each time when doing consecutive searches from a list of queries.

Can I by some magic keep the .2bit genome in memory
Here is my single command

blat -stepSize=5 -repMatch=2253 -minScore=0 -minIdentity=0 \
   ref/hg19.2bit query.dna \
   -ooc=ref/11.ooc \
    out.psl
Loaded 3137161264 letters in 93 sequences
Searched 50 bases in 1 sequences

Thanks in advance

[splaisan]

And the winner is ....
put all queries in a single multifasta.
so simple I did not think about it.

shame on me :-(

[kmcarr]

Another option is to use the gfServer/gfClient programs which are part of BLAT. The gfServer program maintains the genome index in memory and gfClient submits BLAT queries to the gfServer. While the primary use case may be for the server and client to be on different hosts (or one server handling many clients) there is no reason they can not be run on the same computer.

[splaisan]

Thanks @kmcarr, this looks indeed nice too but will require some setup to bring the two entities to life. I should look at this once but my small needs are satisfied with the simplistic multifasta query together with some more stringent filtering to have only one hit per bait.

-stepSize=5 -repMatch=2253 -minIdentity=98 -maxGap=0