A Structure Informed Atlas of Pan-Viral Interactions Reveals Features of Human Infection

While knowledge of protein-protein interactions is critical for understanding virus-host relationships, limitations on the scalability of high-throughput methods have hampered their identification beyond a number of well-studied viruses. Here, we implement an in silico computational framework (termed P-HIPSTer; (Pathogen Host Interactome Prediction using STructurE similaRity) that employs structural information to predict pan viral-human protein-protein interactions. We generate ~280,000 high confidence predictions with an experimental validation rate of ~76%. The resulting inventory identifies specific interactions that play a role in ZIKV and HPV infection and generates novel hypotheses that were functionally interrogated. We demonstrate that P-HIPSTer rediscovers known biology, identifies key regulatory networks mediating virus-human dynamics and reveals novel insights into shared and unique machinery employed across the human virome. Finally, P-HIPSTer reveals a history of evolutionary selective pressure imposed on the human proteome, recapitulating what is known while providing unique insights made possible by the large-scale structure-based P-HIPSTer database