I would like to perform genome-wide meta-analysis of GxE interaction across 5 cohorts.
I'm going to use a joint meta-analysis of Mannig et al. (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3312394/) I know that the method needs SNP beta and SNP×E beta and covariance between the betas.
I could obtained SNP beta and the SNP×E beta from each cohort using PLINK 2.0, but I don't know how to calculate covariance using PLINK 2.0.
Could you tell me how to compute the covariance?