plink 1.9 Principal Component Analysis

Question

[peris]

Hi all,

I have done a plink pca test on my cas/control samples and generated *.eigenval and *.eigenvec files. Now how can I identify the outliers. I am not comfortable in using EIGENSOFT. Is there any other way to do the same.

Thanks and regards.

[ATCG]

Hi, did you find out how to plot the eigenvec? I know you can use plot() but how can one add the sample labels?

Answer 1

[chrchang523]

You should be able to load the .eigenvec file in R, perform clustering on it, and identify outliers in the process.

(The old PLINK way to do this was the --neighbour command. This was based on a basic identity-by-state matrix instead of the variance-normalized GRM, but results should be similar.)

[peris]

Hi chrchang523, thanks for your reply. I need some tip from you again. I know very little R and very very new in GWAS related studies. I can plot grap from the PCA using R; but how can I perform clustering for identifying the outliers.

[chrchang523]

A large number of clustering routines have been written for R. http://cran.r-project.org/web/views/Cluster.html is a reasonable starting point.