Automated docking refinement and virtual compound screening with absolute binding free energy calculations

Abstract

Absolute binding free energy calculations with explicit solvent molecular simulations can provide estimates of protein-ligand affinities, and thus reduce the time and costs needed to find new drug candidates. However, these calculations can be complex to implement and perform. Here, we introduce the software BAT.py, a Python tool that invokes the AMBER simulation package to fully automate the calculation of binding free energies for a protein with a series of ligands. We report encouraging initial test applications of this software both to re-rank docked poses and to estimate overall binding free energies. We also show that it is practical to carry out these calculations cheaply by using graphical processing units in common machines that can be built for this purpose. The combination of automation and low cost allows this procedure to be applied in a relatively high-throughput mode, and thus enables new applications in early-stage drug discovery.Protein-ligand binding free energy calculations based on atomistic molecular simulations promise to play a growing role in drug discovery, as they provide estimates of the binding affinities of compounds proposed as drug candidates for a protein target, and thus may reduce the time and cost required for trial-and-error experimentation. 1,2 Thus, in settings where the calculations are sufficiently fast and accurate, 3 one may anticipate significant savings of time and cost in early stages of drug discovery. [4][5][6][7] It is informative to divide this broad class of methods into two subtypes: relative binding free energy (RBFE) calculations; [8][9][10][11][12][13][14] and absolute binding free energy (ABFE) calculations. [15][16][17][18][19][20][21][22][23] The former (RBFE) estimate the difference in binding free energy between two compounds by computing the change in free energy associated with a non-physical transformation of one compound to the other, in the binding site and in bulk solvent. 8,24 Because it is easiest to carry out such alchemical transformations between compounds that are similar to each other and that adopt similar bound poses, RBFE calculations are often regarded as particularly suitable for the leadoptimization stage of drug discovery, where small chemical modifications of the initial lead compound must be selected. Interestingly, though, a recent study reports greater impact on the earlier hit-to-lead stage. 3

Rigorous Free Energy Simulations in Virtual Screening

“…Interestingly, just before submitting the final version of this Perspective an excellent research paper by Heinzelmann and Gilson entitled, "Automated docking refinement and virtual compound screening with absolute binding free energy calculations", was published. 148 The authors acknowledge that absolute binding free energy calculations can be complex to implement and perform, yet they implemented a fully automated workflow using Python to run AMBER ABFE calculations on a series of ligands. They present encouraging results, both for rescoring docked poses and estimating the absolute binding free energies.…”

Identify potent SARS-CoV-2 main protease inhibitors via accelerated free energy perturbation-based virtual screening of existing drugs

“…For the virtual screening to identify new hits or leads, it is necessary to predict absolute binding free energy (ABFE) for each ligand binding with the target without the requirement to use any reference ligand structure. The FEP-ABFE approach has the advantage of predicting binding affinities between ligands and their targets more accurately than conventional computational methods, such as pharmacophore, molecule docking, and molecular simulations (29)(30)(31). However, the previously used FEP-ABFE calculations are extremely expensive and time consuming and, thus, not suitable for virtual screening purposes (that required to screen a large number of compounds) (32,33).…”

SAMPL7 blind predictions using nonequilibrium alchemical approaches

“…Concerning DSSB in the CD systems, the prediction [50] was done using a bidirectional approach, allowing for the use of the accurate and precise maximumlikelihood Bennett Acceptance Ratio estimator (BAR) [51,52]. Bidirectional estimators in the context of NE approaches are feasible only using strong translational/orientational/conformational volume restraints for the bound decoupled ligand, limiting the accessible conformational space of the system [53] and so yielding reverse and forward work distributions with discernible overlap. The enforcement of conformational restraints, whose contribution to the binding free energy must be somehow assessed by re-weighting procedures, [54] implies the a priori selection of a pose that in real-world drug-receptor systems can be suboptimal leading to systematic (methodological) and unknown errors in spite of an apparent high precision as measured by the inverse of the Fischer information in BAR [52].…”