[adegenet-forum] error in DAPC

[Jombart, Thibaut]

Hi, 

this definitely does not come from the DAPC implementation itself, but possibly kmeans indeed. If you have groups already defined, it would be best use them. K-means is quite fast compared to most other clustering methods used in genetics, so I'm not sure which other approach would work better. I'm guessing some hierarchical clustering methods might be faster.. How long did the clustering step take? If not too long, you could try increasing the max number of steps for kmeans (see ?find.clusters).

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 Mark Coulson [coulsonmw at gmail.com]
Sent: 01 December 2014 12:51
To: adegenet-forum at lists.r-forge.r-project.org
Subject: [adegenet-forum] error in DAPC

Hello,

A while back I asked about the following error when running DAPC

Warning messages:
1: Quick-TRANSfer stage steps exceeded maximum (= 636200)
2: Quick-TRANSfer stage steps exceeded maximum (= 636200)
3: Quick-TRANSfer stage steps exceeded maximum (= 636200)

I got this after running DAPC and choosing my number of PCs to retain as well as the number of groups. It was suggested at the time to be an error in the kmeans implentation. Does this mean I cannot run DAPC on this dataset? It is ~12,000 individuals for a couple of hundred sites at 17 microsatellite markers. I could try sub-setting the data, although the point was to see how these clustered considering them as a whole.

Mark