continuos covariates in Plink

Question

[Laura]

I'm performing a logistic regression in Plink in a gwas data. I have age, sex and race as covariates. Should I use them as dummy categories (age and race)?

Answer 1

[Floris Brenk]

Usually people use them 0 or 1 coded, but this also depends on the number of races you have? For age this depends on the phenotype you have.. For example when you have an age related phenotype like Alzheimer's then you just use age as a continuous covariate. Overall, it might also be good to add the first 6 or 10 (depending on your data) or so MDS dimensions of your genotype data to correct for population stratification.