RAISE: A computational tool for evaluating sarbecovirus spillover potential

Animal sarbecoviruses, relatives of SARS-CoV or SARS-CoV-2, pose a significant zoonotic threat driven by their ability to bind the human ACE2 (hACE2) receptor. To address challenges in evaluating these threats, we developed RAISE (Receptor binding domain-hACE2 Interaction Scoring Evaluation), a computational framework that integrates structural predictions with interaction scoring. By scoring predicted hACE2 interactions, our RAISE model categorized sarbecoviruses into three groups: high potential (hACE2-binding), negligible potential (hACE2-nonbinding), and an intermediate "poised" state (a state defined by either weak binding activity or a high potential to evolve it). Mutation screening of two "hACE2-poised" sarbecoviruses, PDF-2370 and Khosta-1 using RAISE, revealed mutations such as T498Y/W that enabled human ACE2 utilization and expanded their ability to bind to ACE2 receptors from a broader range of species. The model's generalizability was further demonstrated through prospective application to merbecoviruses, highlighting its utility in preemptively assessing zoonotic threats across coronavirus lineages. RAISE provides a predictive roadmap for prioritizing risk viruses and guiding pandemic preparedness.