Inferring epistasis from deep mutational scanning data

Understanding the mutational effects of genetic variants is an important subject in evolutionary biology. Deep mutational scanning (DMS) is a popular mutagenesis method assisting to measure the functional effects of genetic variants at very large scales. However, current state-of-the-art methods for analyzing DMS data can provide inconsistent results across experimental replicates, and the inference of epistasis remains a particular technical challenge. Here, we present a method drawing on previous theoretical advances in population genetics [1] to interpret not only the functional effects of single genetic variants but also the epistasis effects from DMS data. With extensive tests, our analysis reveals more consistent inference of mutational effects across experimental replicates compared to current methods. We also find interpretable epistatic interactions between genetic variants. Our framework can be widely applied to DMS data with multiple generations, replicates, and conditions.