Pose-Rescorer: A Deterministic Single-Frame MM/GBSA Workflow for Post-Docking Ligand Ranking

Abstract Molecular docking is widely used for large-scale virtual screening, yet its empirical scoring functions are approximate and often fail to capture the balance between intermolecular interactions and solvation. Molecular dynamics–based free-energy methods can address these limitations but are computationally expensive, non-deterministic, and impractical for routine post-docking prioritization. Here, we present Pose-Rescorer, a chemistry-first, deterministic post-docking rescoring tool that applies single-frame MM/GBSA or MM/PBSA calculations to minimized protein–ligand complexes for relative pose and ligand ranking. Pose-Rescorer enforces strict chemical correctness by requiring explicit ligand connectivity, GAFF2 parameterization with AM1-BCC charges, and a fixed receptor context, ensuring internal comparability of scores. The workflow performs restrained generalized Born minimization followed by single-structure MM/GBSA or MM/PBSA evaluation without molecular dynamics, conformational sampling, or entropy estimation; consequently, scores are not interpreted as physical binding free energies but are intended solely for relative ranking within a consistent receptor system. To complement single-frame rescoring, Pose-Rescorer optionally incorporates Rapid Perturbation Sampling (RPS), a reproducible diagnostic procedure that quantifies the numerical sensitivity of MM/GBSA scores to small coordinate perturbations. Worked examples using kinase, protease, bromodomain, and chaperone systems demonstrate how Pose-Rescorer refines docking-derived ligand prioritization. Pose-Rescorer is freely available as open-source software at https://github.com/Amirtesh/Pose-Rescorer.