[ssml] Genomics & RAM

[Robert Citek]

On Tuesday, January 6, 2004, at 10:29  PM, Tristan Fiedler wrote:
> I am interested in setting up a bioinformatics machine to do in-house
> blast & clustering with 10's to 100,000's of ESTS.
>
> How much RAM is needed to make these tasks bearable?  Any experiences
> welcome.

That depends on a number of factors, including what you consider 
"bearable".  More is definitely better because you want to avoid having 
to swap to disk.   If you are running blast on a UNIX-like machine 
(e.g. linux), I would recommend that you monitor a few blasts with top 
and vmstat.  If you discover that you are running out of memory and 
having to swap to disk during your blasts, then you should consider 
getting more memory.

That's just to start.  For more details on choosing hardware for 
running blast, see the O'Reilly book ( 
http://www.oreilly.com/catalog/blast/index.html ).  Not only do the 
authors go into details about hardware, but they also give excellent 
advice on various parameters you can tweak to speed up the blast.

Hope this helps and let us know how things go.

Regards,
- Robert