[adegenet-forum] find.clusters reproducibility

[Jombart, Thibaut]

Hi there, 

yes, it is not an unusual problem. Usually it is not a 'convergence' (strictly speaking) issue (so n.iter does not matter to much) but more that different starting points give different optimal clusters. Try increasing n.start to 200? 

Cheers

Thibaut
________________________________________
From: adegenet-forum-bounces at lists.r-forge.r-project.org [adegenet-forum-bounces at lists.r-forge.r-project.org] on behalf of Catarina Pinho [catarina at cibio.up.pt]
Sent: 10 October 2012 14:33
To: adegenet-forum at lists.r-forge.r-project.org
Subject: [adegenet-forum] find.clusters reproducibility

Hello

I am totally new to adegenet and I am finding it very useful.

I have a data set comprising 710 individuals and 22 loci in which BAPS detected 49 clusters. I was going to try to use DAPC on these data so I began with the find.clusters function.

I have ran the function multiple times choosing different Ks (BIC plot is flat around 4-8 clusters and I tried with all these values). I also tested multiple times with the same K to test for reproducibility. I get different partitions of the individuals into clusters according to the run. I tried to increase n.iter but the problem persists (and over 1e9 the function outputs an error).

Can you help me?

Thank you.

Catarina