AToM: An open-source software package for relative alchemical binding free energy estimation in structure-based drug design

Molecular dynamics-based relative alchemical binding free energy computations are one of the work-horses of modern structure-based drug design campaigns in academic and industrial settings. However, in addition to the inherent chemical complexity of the macromolecular targets, successful deployments remain challenging due to the limitations and poor accessibility of software tools. Academic open-source packages, in particular, require specialized expertise to set up and generally do not reliably support scaffold-hopping and charge-changing transformations that are often encountered in virtual screening applications. Often, even advanced atom-mapping algorithms fail to identify suitable transformation paths between ligand pairs. Implementing single and dual-topology alchemical protocols at all levels is challenging due to the extensive customizations of molecular dynamics engines and simulation topologies required. Our group has recently developed the Alchemical Transfer Method (ATM) that addresses many of these challenges. The method is based on a coordinate transformation that directly connects the bound and unbound states of the two ligands of the target ligand pair. It employs a single simulation system prepared in a standard topology using conventional tools. Because it does not require atom-mapping and soft-core modifications, ATM can be used with any force field and conformational sampling algorithm available in the underlying MD engine. The AToM software implements ATM with the popular OpenMM GPU MD engine. I will present the foundations of the method, go over a typical setup and simulation workflow, and show recent applications of the AToM software to academic and proprietary datasets.