ADMIXUTRE cross validation (cv)

Question

[simone.castellana]

Hi everyone,

I’d need some support for ADMIXTURE, specifically for the cross validation procedure. My problem is that the resulting CV error produces extremely low values, so there’s must be something wrong with my input (e.g. CV error (K=2): 0.09725)

My input file consists of 934 samples, that I have previously filtered using bcftools for having a calling rate >95% and only biallelic sites, using this command:

bcftools view -i ‘F_MISSING<0.05’ -m2 -M2 -v snps myvcf.vcf.gz -Oz -o myfilteredoutput.vcf.gz

My resulting file has circa 4million SNPs.

So, after converting the vcf file into bed using Plink, I run admixture, like so:

admixture --cv myfilteredoutput.bed 2 > log2.out 

(for K=2 in this example)

What I’m doing wrong?

Thanks