Improving docking and virtual screening performance using AlphaFold2 multi-state modelling for kinases

                  The Paper of the Day discusses the use of multi-state modeling (MSM) with AlphaFold2 (AF2) to improve structure-based virtual screening (SBVS) performance for kinase targets. Kinases are important drug targets that exhibit diverse conformational states, but experimental structures are biased towards the active DFGin state. The standard AF2 protocol also tends to produce kinase structures in the DFGin state, limiting the discovery of diverse kinase inhibitors. The authors developed an MSM protocol that uses state-specific structural templates to guide AF2 in modeling kinase structures in different conformations. The MSM models showed comparable or improved structural accuracy compared to standard AF2 models. In cognate docking experiments, the MSM models exhibited enhanced binding pose prediction accuracy over standard AF2 and AlphaFold3 models. For virtual screening, the ensemble docking using MSM models consistently outperformed standard AF2 and AlphaFold3, particularly in identifying diverse hit compounds. The authors found that the MSM approach was more advantageous when the active compounds in the screening library were more diverse. Overall, the study demonstrates the potential of MSM in broadening kinase inhibitor discovery by facilitating the identification of chemically diverse inhibitors.