[mira_talk] Re: Preprocess memory requirement
- From: Bastien Chevreux <bach@xxxxxxxxxxxx>
- To: mira_talk@xxxxxxxxxxxxx
- Date: Thu, 2 Apr 2015 21:17:55 +0200
On 02 Apr 2015, at 4:59 , Torben Nielsen <torben@xxxxxxxxxx> wrote:
If I am going to run MIRA to just pre-process, is there a good way of working
out the memory requirement?
Sorry, no. Some parts are pretty predictable, like the amount of memory needed
by the pure read data alone. Many of the preprocessing steps are also not
memory intensive, all but one: kmer counting for -CL:pec (and a few other
things nowadays).
As soon as you are going into kmer counting, things become unpredictable and
are dependent on the coverage of your data. The initial counting implemented in
MIRA is similar (but much simpler) to what was published a few years later as
DSK (disk streaming of kmers). Low coverage data create larger files on disk
which, in a last step, need to be loaded back into memory as a single chunk and
that can get quite big (unpredictably so).
B.
--
You have received this mail because you are subscribed to the mira_talk mailing
list. For information on how to subscribe or unsubscribe, please visit
http://www.chevreux.org/mira_mailinglists.html
Other related posts:
