DUckCov: a Dynamic Undocking‐based Virtual Screening Protocol for Covalent Binders

Moira Rachman Andrea Scarpino Dávid Bajusz Gyula Palfy Istvan Vida Andras Perczel Xavier Barril György M Keseru

ChemMedChem, 2019. doi:10.1002/cmdc.201900078

Thanks to recent guidelines, the design of safe and effective covalent drugs has gained significant interest. Other than targeting non‐conserved nucleophilic residues, optimizing the non‐covalent binding framework is important to improve potency and selectivity of covalent binders towards the desired target. Strong efforts have been made in extending the computational toolkits to include a covalent mechanism of protein targeting, like in the development of covalent docking methods for binding mode prediction. To highlight the value of the non‐covalent complex in the covalent binding process, here we describe a new protocol utilizing tethered and constrained docking in combination with Dynamic Undocking (DUck) as a tool to privilege strong protein binders for the identification of novel covalent inhibitors. At the end of the protocol, dedicated covalent docking methods were used to rank and select the virtual hits based on the predicted binding mode. By validating the method on JAK3 and KRas, we demonstrate how this fast iterative protocol could be applied to explore a wide chemical space and identify potent targeted covalent inhibitors.