Do the covariates included in a GWAS vary across different traits? How should we determine the most appropriate covariates for a GWAS analysis?
Yes - Be guided by the phenotypic/epidemiological work in the trait BUT avoid including covariates that are heritable
There are some basic covariates you always want to have (like your genetic PCs). The others will depend on your knowledge of the phenotype and what might affect the relationship between the exposure and the outcome.
And as Sarah mentioned, you don’t want to adjust for a covariate that is heritable (like BMI) - here’s a paper that talks about why: A Robust Example of Collider Bias in a Genetic Association Study - PMC