Variation in the genome is a powerful instrument for learning about differences between individuals. It avoids the problems of reverse causality, is abundant in natural populations, and can be generated in a targeted manner in the lab. I will talk about our approaches for using different types of genetic variation to understand cellular traits, and the computational models we developed for the purpose. We previously traced the signal from common alleles to RNA abundance and protein levels, as well as cellular growth rate, finding both shared and independent causes of variability. Currently, we employ genome engineering methods to learn about gene essentiality. To undertake experiments at scale, we have developed models to predict the mutational outcome of CRISPR/Cas9 editing, and to efficiently analyse genome-wide screens. I will present the data motivating these models, their formulation, inference, and results.ge