[adegenet-forum] xval warning

Jombart, Thibaut t.jombart at imperial.ac.uk
Mon Nov 9 15:59:29 CET 2015


Hi there,

this is not an error but a warning, which in this case is most likely harmless. I posted an issue for it:
https://github.com/thibautjombart/adegenet/issues/103

However with 4 individuals in your smallest cluster I'm afraid cross validation is not really practicable, and I would not trust the results.

Missing data are replaced by the mean value of the variable. I suspect you use a different method for replacement of missing values, and/or use (implicitly, probably), a different scaling option when passing different types of data to the function.

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 LAURA NICOLE WOODINGS [17869067 at students.latrobe.edu.au]
Sent: 01 November 2015 02:22
To: adegenet-forum at lists.r-forge.r-project.org
Subject: [adegenet-forum] xval warning

Hi,

I am trying to use xval to cross-validate the number of PCs to retain for a DAPC using the below command and I am getting the following error msg:

xval_nopopbp75_cnt75_cl96_maf5to1 <- xvalDapc(nopop_bp75_cnt75_cl96_maf5to1NoNA at tab, nopop_clust$grp)

There were 50 or more warnings (use warnings() to see the first 50)
> warnings()
Warning messages:
1: In if (result == "overall") { ... :
  the condition has length > 1 and only the first element will be used
2: In if (result == "groupMean") { ... :
  the condition has length > 1 and only the first element will be used

Using find.clusters 2 groups are found but 1 group has 83 individuals and the other group has 4. Is the discrepancy in sample size causing the problem with xval? And if so how can I determine the optimal number of PC’s to keep? I have attached my input file and the script that I use.

Also when conduction DAPC should you remove the NA’s in the data? The DAPC seems to work when the Nas are not removed, but when I run it with them removed I get different results.

Thanks for your help!

Cheers,
Laura

PhD Candidate
Department of Ecology, Environment and Evolution| La Trobe University | Bundoora 3086 | Australia

m: +61 408 642 006 | e: 17869067 at students.latrobe.edu.au
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.r-forge.r-project.org/pipermail/adegenet-forum/attachments/20151109/f5ab732b/attachment.html>


More information about the adegenet-forum mailing list