Research | Ray Luo Lab

Polarizable Gaussian Multipole force field

Fri, 05 Jun 2026 00:00:00 +0000

We develop the polarizable Gaussian Multipole (pGM) model — a physically motivated force field in which atomic charges and multipoles are represented by smooth Gaussian distributions. Our work spans electrostatic parameterization (PyRESP, PCMRESP), refinement of atomic polarizabilities, energy-conserving induced-dipole schemes, periodic-boundary electrostatics via the isotropic periodic sum, and stress/pressure control for accurate condensed-phase molecular dynamics.

Machine learning for molecular surfaces and implicit solvent

Thu, 04 Jun 2026 00:00:00 +0000

We bring machine learning to implicit-solvent modeling: graph and convolutional neural networks that learn the solvent-excluded molecular surface, and data-driven models of Poisson–Boltzmann reaction-field energies. Implemented on GPUs (AmberTorchPB), these methods make continuum electrostatics fast and smooth enough for routine biomolecular simulation.

Binding free energies — MM/PBSA and alchemical methods

Wed, 03 Jun 2026 00:00:00 +0000

We develop methods to predict binding free energies for drug discovery: GPU-accelerated MM/PBSA for protein–ligand and membrane-protein systems, and alchemical free-energy perturbation made robust by singularity-free softcore potentials (DEGAUSS). The goal is accurate, efficient affinity prediction across diverse targets.

Allostery and disease-relevant protein dynamics

Tue, 02 Jun 2026 00:00:00 +0000

We use molecular dynamics and free-energy analysis to uncover cryptic allosteric sites and to interpret disease-associated mutations in signaling proteins and cancer targets (for example SHP2 and the Hippo-pathway kinases), connecting conformational dynamics to function and therapeutic opportunity.