LMM GWAS - why are population covariates included?

Question

[ucbtep]

Hi! I'm new to GWAS and have a theoretical question:

If LMM-based GWAS software (e.g., regenie) can handle sample relatedness (population structure & cryptic relatedness), why are the population covariates included in the analyses?

Thanks for your help :)

[4galaxy77]

Good question - this is something I asked a colleague just yesterday and I believe the answer was that including PCs as covariates helps speed up the estimation of the null GRM used to sample relatedness. But I will double check this and come back..