Sully Research Group

Method

The Shared Spine

Every pillar runs on the same closed loop: quantum → force fields → molecular dynamics → AI → experiment → back.

1. Quantum calculations2. Force fields + MDpredicted vs measured3. AI selects the next experimentMass spectrometer4. Real measurementexample: Cannflavin AThe gap between prediction and measurementbecomes the next question.

Core expertise

Molecular dynamics

Simulations are only as good as the forces behind them. Drag the sliders to reshape the Lennard-Jones potential.

r (Å)V(r)attraction (van der Waals)repulsion
ε (well depth)1.00
σ (contact distance, Å)3.40

V(r) = 4ε [ (σ/r)¹² − (σ/r)⁶ ]

Two atoms approach while the potential curve tracks their energy.

Quantum scans to parameters

dimer scan: distance + angleQM energy points · LJ fitMonte Carlo · Bayesian · particle swarm · ant colony
Quantum potential-energy scans of mixed chemical species on ~2,000 CPUs, used to optimize CHARMM General and Drude force-field Lennard-Jones parameters.

Simulation gallery

Fuel mixtures

Computational

Predicting thermodynamic properties of liquid fuel blends (OpenMM, CHARMM, GPU clusters).

Supercritical water

Computational

Molecular dynamics of supercritical water for nuclear-reactor coolants, with the Kharkiv Institute of Physics and Technology.

Cannabinoid solubility

Computational

Automated screening of temperature-dependent cannabinoid solubility in industrial solvents (manuscript in preparation).

Protein–ligand dynamics

Computational

Plant compounds in kinase binding pockets. Links to the Plants pillar.

Video coming soon

Why van der Waals parameters matter for every simulation

Video coming soon

Molecular dynamics trajectory reel