Covariates for ML-corrected DRAGEN pVCF

Hi all!

are there any recommendations for which if any covariates to use for regenie tests with ML-corrected DRAGEN pVCF? (In addition to other standard covariates like sex, age, age squared, assessment center, PCs).

Using sequencing provider (Field 32051) as batch seems like an obvious choice, though the effect from sequencing are not as clear to me as for example exome capture has on WES data. I could not find a paper that already uses 500k WGS to compare.

I appreciate any tips!

Thanks in advance