As a team, we are curious about how the constant jiggling of atoms and the interactions between them give rise to complex biomolecular behavior. How do biomolecules communicate with each other, respond to external stimuli, or sense their environment? With modern computational resources, can we reliably simulate these living machines across larger length scales and longer time scales than currently possible? Can we decipher the underlying principles of life at the molecular level and use this knowledge to develop new strategies for understanding and treating disease? To address these questions, we combine theoretical chemistry, computational modeling, and artificial intelligence to accelerate discovery and deepen mechanistic understanding.

AI-Augmented Molecular Simulations

We develop AI-driven methods to accelerate and enhance molecular simulations, with a focus on rare events and long-timescale dynamics.
By combining enhanced sampling with modern machine learning, we explore new ways to push molecular simulations into regimes that were previously out of reach.
For example, can we reimagine the protein folding problem as a game, and train an AI agent that learns how to navigate its energy landscape efficiently?

Causal Discovery for Biomolecular Mechanisms

Biomolecular systems are rich in correlations—but not all correlations reflect true mechanisms.
Our group uses causal discovery to understand how local conformational changes propagate through a molecule and ultimately give rise to global function. By integrating molecular dynamics with causal inference, we seek to go beyond pattern recognition and identify relationships that are truly mechanistic.
The goal is to turn high-dimensional simulation data into clear, testable hypotheses that can guide and inform experiments.

Computational-Driven Drug Discovery

Drug discovery is often framed as a search problem. But what if it is fundamentally a problem of understanding?
If we truly understand the mechanisms by which molecules interact with their targets, can we move from empirical screening to rational design? Our group explores how mechanistic insight, gained from physics-based simulations and machine learning, can guide more reliable and principled drug discovery.
A natural question we ask is whether deeper understanding can translate into more efficient development pipelines and ultimately, more accessible and affordable medicines.