Exploring Large Combinatorial Chemical Spaces to Accelerate Drug Discovery with Multisite λ-DynamicsSpeaker: Jonah Vilseck, PhD, Assistant Professor, Department of Biochemistry & Molecular Biology, IUSM Location: 402 N Blackford St. Indianapolis, IN 46202 LD 010
Free energy differences underlie a variety of biologically relevant processes, and as such, their computation can provide valuable insight into solvation, association, and conformational equilibria. Rigorous, physics based alchemical free energy calculations have become a powerful computational tool to study these processes and perform molecular design. Traditional methods, such as free energy perturbation theory or thermodynamic integration (TI), have been widely developed and employed for these purposes, but generally suffer from an inherent scalability limitation and an additive expense of running stratified simulations at discrete physical or alchemical states of a biochemical system. In contrast, by utilizing a continuous alchemical coupling parameter, λ-dynamics can successfully overcome these disadvantages. Multisite λ-dynamics (MSλD) can investigate many different substituent modifications at one or multiple sites around the common core of a molecule simultaneously within a single molecular dynamics simulation. Thus, tens to hundreds of relative free energy differences can be readily computed and significant resource cost savings are achieved. This work will demonstrate the power of MSλD to explore combinatorial chemical spaces within the context of a pharmaceutical design problem. We find no loss of statistical precision in computed free energies of binding with MSλD compared to conventional TI and multistate Bennett acceptance ratio (TI/MBAR) calculations; yet MSλD is at least 8–20 times more efficient. Excellent agreement with experimental IC50s is also observed, with mean unsigned errors of 0.5–1.0 kcal/mol. The combination of high accuracy and precision in computed free energy differences with the ability to explore large chemical spaces makes MSλD an attractive alternative to conventional free energy methods for a variety of applications, especially computer-aided drug design.