[adegenet-forum] Snapclust for individual assignment

Thibaut Jombart thibautjombart at gmail.com
Mon Mar 5 17:37:39 CET 2018

Hi Adam,

to predict group memberships probabilities for a pre-defined group without
optimization, you need to:
- set the prior to the group you're interested in
- run snapclust with zero iterations: max.iter = 0

For instance:

snapclust(sim2pop, k = 2, pop.ini = pop(sim2pop), max.iter = 0)

If that is not a satisfying workaround, then it is best to post this as an
issue on github; I can't guarantee I'll have time for this in the near
future, but it isn't very difficult to implement and someone else might
give it a shot.


Dr Thibaut Jombart
Lecturer, Department of Infectious Disease Epidemiology, Imperial College
Head of RECON: repidemicsconsortium.org
WHO Consultant - outbreak analysis
Twitter: @TeebzR
+44(0)20 7594 3658

On 5 March 2018 at 14:51, Adam Andrews <adam.andrews at uit.no> wrote:

> Hi,
> I would like to use Snapclust to obtain group membership probabilities for
> unknown individuals in respect to pre-defined reference populations.
> As the reference populations are already established, I would
> like to define the reference clusters *a priori*​, then define k = the
> number of reference clusters, and the individuals to assign as an open
> group so to speak. However, k must match the number of priors and I can see
> no way to provide priors for the reference samples and not the individuals
> to assign. Is there a way around this?
> This seems a slightly different approach to what you have documented
> - using find.clusters and so I just wanted to check if this is a
> possibility.
> Best,
> Adam Andrews
> Researcher
> UiT The Arctic University of Norway
> _______________________________________________
> adegenet-forum mailing list
> adegenet-forum at lists.r-forge.r-project.org
> https://lists.r-forge.r-project.org/cgi-bin/mailman/
> listinfo/adegenet-forum
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.r-forge.r-project.org/pipermail/adegenet-forum/attachments/20180305/3b0be9d3/attachment.html>

More information about the adegenet-forum mailing list