Genotype PCA from genotype matrix

Question

[i.sudbery]

If I have a individual x locus genotype matrix, what is the easiest way to calculate genotype principal components? Can I just use prcomp in R?

This lovely post Produce PCA bi-plot for 1000 Genomes Phase III - Version 2 goes through the process of generating a PCA from VCF files, but I wondered if I needed to use PLINK if I already had the matrix?

[Kevin Blighe]

I still don't know if PLINK, GEMMA, SNPRelate, etc. are doing anything other than generating a standard covariance matrix from the genotypes, and then just using that. Similar question here: https://bioinformatics.stackexchange.com/questions/4030/pca-on-genotype-matrix-with-multiple-alleles

Answer 1

[i.sudbery]

I don't know if this is absolutely the most optimal way to do it, but we did straightforward prcomp on the genotype matrix and the first 2 PCs more or less recapitulated what we know about the population structure of the samples.