Energy sampling against biomolecular ensembles has been a core research area for decades, and recent advances - in both hardware and software - have pushed the accessibility of these methods into the mainstream. The multi-dimensional integral of the partition function is analytically inaccessible which has led to the development and application of numerical approximations for molecular ensemble sampling and partition function calculation. Conventional molecular dynamics (MD) driven free energy methods are generally CPU/GPU-intensive due in large part to sampling convergence difficulties given the complicated energy landscape observed in protein:ligand binding. MovableType (MT) addresses this sampling problem through fast local partition function estimation using one or more initial molecular conformations coupled with pair potential sampling and recombination (a process referred to as local sampling). Given a generated ensemble of protein:ligand poses (e.g., global sampling), MT provides extended local sampling of that energy state ensemble. This method has shown itself to be competitive with other approaches in the prediction of the free energy of ligand binding, but at much lower computational cost. These ensemble poses are often generated using conventional or MT-based docking, but in cases in which docking poses are inaccurate or highly strained, success can be improved with ensembles generated using other methods including direct structure manipulation (e.g., rotamer, loop, and protonation state sampling); multiple NMR, X-ray, or cryogenic electron microscopy (Cryo-EM) experimental models; and/or molecular dynamics or Monte Carlo simulation trajectories. Recently, we explored the compatibility of the MT method with protein:ligand complex poses from different global sampling methods and we validated these results against several structurally and mechanistically diverse benchmark sets. In this talk, we will discuss the impact of these sampling methods on our predictions and explore how they are deployed in routine drug discovery campaigns.

Rigorous free energy calculations using MovableType with molecular dynamics driven protein: Ligand sampling

ACS Fall 2021

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.