Joint estimation of gene-by-context covariance across many exposures in biobanks

Genetic effects on complex traits change with age, sex, medication, and lifestyle, and these contexts are correlated, so one-exposure-at-a-time analyses cannot tell which context modifies genetic effects. This work jointly estimates the covariance of per-variant genetic effects across many contexts at biobank scale and separates amplification, a shared rescaling of genetic effects, from effect heterogeneity, with applications to UK Biobank traits.