Proteins function as dynamic ensembles of interconverting conformations shaped by free energy landscapes. I combine computational and experimental biophysics with machine learning to understand how protein dynamics encode function, how biological perturbations reshape conformational landscapes, and how these principles can be used to design new proteins with controllable activities.
Pushing HX-MS to residue resolution
Hydrogen exchange-mass spectrometry (HX-MS) is uniquely suited to probe protein dynamics in solution under near-physiological conditions, but its readout has traditionally been limited to peptide-level resolution. I develop computational and experimental methods to extract residue-level information from HX-MS data using Bayesian inference and machine-learning models. These methods make it possible to detect subtle, site-specific changes in dynamics that are invisible to static structures.
Mapping conformational landscapes
Allostery — communication between distant sites within a protein — is one of the most pervasive yet poorly understood mechanisms in biology. I combine high-resolution HX-MS methods, molecular dynamics simulations, and machine-learning-based analysis to characterize how ligand binding, mutations, and other biological perturbations reshape the conformational landscapes of allosteric systems.
Designing protein dynamics de novo
My long-term goal is to use the dynamic principles uncovered above to design allosteric proteins with controllable functions. By coupling generative protein design with explicit modeling of conformational ensembles, I aim to engineer proteins whose activity can be switched by small molecules, light, or other inputs — providing both new tools for synthetic biology and new ways to test our understanding of how protein dynamics are encoded in sequence.