Parallel calculation in dadi

143 views
Skip to first unread message

吴琦

unread,
Dec 3, 2010, 7:25:35 AM12/3/10
to dadi...@googlegroups.com
I want to work a very large data set with dadi which may include several million SNPs. It will be two slow for any single CPU to work. So I wonder whether dadi can perform multi processes mission. Therefore the calculation speed could be increased. I note that there is a RunInParallel.py script file in dadi package, but it seems it is not fully tested and not mentioned in manual file.

WQ

Ryan Gutenkunst

unread,
Dec 3, 2010, 1:11:40 PM12/3/10
to dadi...@googlegroups.com
Hello WQ,

First, the computational speed of dadi depends on the number of
populations and the number of samples from each population. The total
number of SNPs doesn't matter. So several million SNPs is no problem.

Second, right now dadi is not really parallelized. Some aspects of the
optimization could be, if that becomes a bottleneck for you. I suggest
you try your problem first, to see whether heavy parallelization will
be worthwhile.

Best,
Ryan

> --
> You received this message because you are subscribed to the Google Groups
> "dadi-user" group.
> To post to this group, send email to dadi...@googlegroups.com.
> To unsubscribe from this group, send email to
> dadi-user+...@googlegroups.com.
> For more options, visit this group at
> http://groups.google.com/group/dadi-user?hl=en.
>

--
Ryan Gutenkunst
Assistant Professor
Molecular and Cellular Biology
University of Arizona
phone: (520)626-0569
http://gutengroup.mcb.arizona.edu

Reply all
Reply to author
Forward
0 new messages