We released a preprint on VPA and Li molecular quantitative trait loci in primary human neural progenitors. Gene by treatment (GxT) interactions are critical for understanding psychiatric disorder risk and variability in treatment response, yet their discovery remains challenging due to the complexity of human exposures. For example, while epidemiological birth cohorts demonstrated that valproic acid (VPA) taken during pregnancy strongly increases the likelihood of an autism spectrum disorder (ASD) diagnosis in the exposed child, only a small proportion of those exposed are affected with ASD. Similarly, while lithium has been successfully used as a treatment for bipolar disorder for over 75 years, around half of individuals treated have a beneficial clinical response. It is likely that genetic variation plays a strong role in response to each treatment, but pharmacogenomic studies are often underpowered and confounded by poorly controlled concentration, duration, and adherence of treatment. Here, we demonstrate a novel approach to dissect these interactions by investigating how common genetic variation alters gene regulatory responses to clinically relevant perturbations, which we call “GxT in a dish”, where the molecular response of each donor to well-controlled exposures can be precisely measured.