Individual human genomes frequently contain variants that have epistatic interactions

Coevolution has been widely studied between species and has an important role in our understanding of biological function. For proteins there has recently been interest in the identification of positions within proteins that have coevolved including their use for modelling protein structure. Such studies focus on the identification of coevolving positions (or evolutionary couplings) within multiple sequence alignments of proteins from many species. Here, we exploit large human genome resources to investigate if it is possible to use genetic variation data from a single species, human, to identify positions within proteins that have coevolved. We combine the 1000 genome project genetic variation data with protein structural data to identify variant-variant interactions within individual human genomes. We find >4,000 combinations of variants that are located close in 3D protein structure and >1,200 in protein-protein interfaces. Many variant combinations include compensatory amino acid changes (e.g. maintaining charge/functional groups), thus supporting that these are coevolutionary events. Our study highlights that it is possible to identify coevolution within a single species. Given the important role that genetic variation plays in causing disease it is important for variant interpretation and precision medicine to consider the gestalt effects of variants rather than individual variants in isolation.