#31 · Structural bio & protein design
DFT reference data refines torsion parameters and hybrid ML/MM simulation setup
The dominant thread is quantum-mechanical reference data supplying the fidelity of molecular energy models for biomolecular simulation. AFFDO, a GPU-accelerated platform, refits GAFF2 torsion parameters for drug-like ligands against DFT reference data, improving agreement with QM. CHARMM-GUI Hybrid ML/MM Builder automates preparation of protein-ligand systems and input files for hybrid neural network potential/molecular mechanics simulations. What is new is the coupling of GPU-accelerated DFT-level reference data with parameter refitting and hybrid ML/MM system construction. The DBFE preprint, an alchemy-free absolute binding free energy method in implicit solvent with publicly released code, addresses a separate problem and is not representative of this thread.
Published papers
CHARMM-GUI Hybrid ML/MM Builder for Hybrid Machine Learning and Molecular Mechanical Modeling and Simulations.
CHARMM-GUI Hybrid ML/MM Builder automates preparation of protein-ligand system and input files for hybrid neural network potential/molecular mechanics…
Automated Force Field Developer and Optimizer Platform: Torsion Reparameterization.
AFFDO, a GPU-accelerated automated platform that refits GAFF2 torsion parameters for drug-like ligands against DFT reference data, improving QM agreement and…
Preprints
Binding Free Energies without Alchemy
Proposes Direct Binding Free Energy (DBFE), an alchemy-free end-state absolute binding free energy method in implicit solvent, with code released publicly.