Molecular Dynamics and Artificial Intelligence Based Tools to Innovate Covalent and Noncovalent Drug Design

Abstract Cancer and infectious diseases are threatening modern human kind. To combat the challenge, targeted cova- lent inhibition and targeted protein degradation are emerging as the new frontiers in drug discovery. This SBIR project seeks to meet two unmet needs in the emerging covalent as well as traditional reversible drug discovery programs by developing two commercial solutions based on cutting-edge research in molecular simulations and data science. In the Phase I period, ComputChem LLC has developed a local/cloud software application iTitrate that can offer accurate and reliable predictions of protein pKa values and nucleophilic cysteine and lysine. As the first of its kind, iTitrate is based on a unique and extensively validated constant pH molecular dynamics approach and a very accurate and efficient generalized Born implicit-solvent model. In Phase I, ComputChem also devel- oped a prototype of a novel small-molecule pKa prediction tool based on artificial intelligence. Building on the Phase I progress, the objective of the Phase II project is to significantly accelerate, enhance, and demonstrate the capabilities of iTitrate and improve the accuracy of iKa. iTitrate can be applied to expedite efforts in a vari- ety of covalent drug discovery programs, including targeted covalent inhibitor design; kinome or proteome wide scanning for covalently targetable sites; protein-protein interaction inhibition; and proteolysis targeting chimera design. iTitrate can be further developed and integrated with iKa to offer a novel solution for accurate protonation state assignment and pH-dependent modeling of protein-ligand complexes, which would add significant value to computer-aided lead optimization and other in silico drug design tasks.