Answers are grounded in this paper and cite their sources.
Evaluates a ligand-protein affinity protocol in OpenMM, indicating reliable calibration methods for natural products.
Correction (new version, 2026-07-21) — integrity correction. Correction (version 2, 2026-07-21) - fabricated ligand-panel annotations. After this record was first deposited (2026-05-04), a primary-source audit (2026-07-16) established that the 15-ligand calibration panel used in Section 3.6 (data/chembl_mmp1_calibration.csv) carries fabricated compound names, potency values, and literature attributions. A structure-first PubChem lookup finds that all seven entries naming a specific drug are a different molecule than named - e.g. the entry labelled prinomastat is C23H30N2O5S against prinomastat's C18H21N3O5S2 (PubChem CID 466151) - and 14 of the 15 structures are unknown to PubChem (~119 M compounds). No script in the repository generates the file and it carries no retrieval record; a previous automated session produced it and the circumstances are not recoverable. This version withdraws Section 3.6 in full (the Boltz-2-vs-pIC50 calibration, being a correlation against fabricated potency, has no verified ground truth) and removes every compound-identity and potency claim derived from the panel. Unaffected: the absolute binding free energy pipeline itself, its T4 lysozyme L99A / benzene benchmark, and the EMB-3 application - none use the panel. Full scope: preprints/_metadata/FABRICATED_PANEL_SCOPE_2026_07_16.md. Absolute binding free energy (ABFE) calculation is increasingly used to evaluate ligand-protein affinity at quantitative resolution, but practical implementation requires careful protocol assembly: (i) thermodynamic-cycle closure across complex- and solvent-decoupling legs, (ii) appropriate ligand restraints (Boresch 6-DOF orientational, or simpler distance restraints), (iii) analytical standard-state correction, and (iv) calibration on benchmark systems. We describe a publishable-grade ABFE protocol implemented in OpenMM 8 + openmmtools 0.26, using a flat-bottom centroid distance restraint between ligand and binding-site receptor anchors (k = 10 kcal mol⁻¹ Å⁻², r_max = 8 Å) and the analytical standard-state correction ΔG_R° = -RT ln(V_R / V°), where V_R = (4/3)π r_max³ and V° = 1660.5 ų. The cycle assembles as ΔG_bind = ΔG_solvent_decouple − ΔG_complex_decouple − ΔG_R°. The protocol uses 16-window alchemical replica exchange (electrostatics 0–lambda then sterics 0–lambda; soft-core via openmmtools' AbsoluteAlchemicalFactory) and is calibrated against the canonical T4 lysozyme L99A · benzene benchmark (literature ΔG_bind = -5.18 ± 0.18 kcal/mol, Mobley et al. JCP 2007). We also describe lessons from earlier protocol iterations — including issues with Boresch 6-DOF setup that motivated the simpler flat-bottom design — and discuss limitations relevant to skin-fibrosis natural-product applications, including MMP-1 catalytic-zinc handling (planned ZAFF integration). Complete code is available open-source under Apache-2.0. Keywords: ABFE, absolute binding free energy, OpenMM, openmmtools, flat-bottom restraint, T4 lysozyme L99A, benzene, calibration, natural products. ---
No takes yet. Share an insight, caveat, or question.
Cheongwoo Han (2026) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: