Research

I investigate the physical principles by which protein scaffolds control catalytic outcomes using hybrid QM/MM simulations (DFT, DLPNO-CCSD(T)), microsecond-scale molecular dynamics, and dynamic 3D electric field mapping. A central focus has been the non-heme iron/2OG enzyme superfamily — where I have uncovered how coordination dynamics, substrate reorientation, and second coordination sphere interactions drive catalytic diversity, including predicting a novel bicarbonate intermediate validated in Science (2021). I have mapped three-dimensional electric fields within active sites as quantitative descriptors of catalytic potential, revealing latent compatibility with non-native functions — a physical basis for evolving new chemistry. Using microsecond MD, I characterized allosteric communication pathways and correlated motions that modulate catalysis far from the active site, and developed validated force field parameters for metalloenzyme simulations.

I develop computational methods, open-source software, and machine learning approaches that translate mechanistic understanding into predictive power and enzyme design. Applying ML to electric field fingerprints, I showed that models can classify enzymatic function with up to 84% accuracy — revealing electrostatic signatures as the physical basis of functional divergence across enzyme families and enabling the prediction of evolutionary starting points for new reactivity. This work led to PyCPET, an open-source toolbox for computing heterogeneous 3D protein electric fields, integrating MD, topological analysis, and QM/MM into a unified pipeline. I demonstrated that directed evolution optimizes the enzyme’s electric field — not just individual residues — in protoglobin’s non-native cyclopropanation activity, establishing the second coordination sphere as a powerful lever for rational design. My long-term vision is inverse enzyme design: starting from a desired catalytic outcome, identifying the optimal electric field, and using ML with AlphaFold/diffusion models to generate scaffolds that produce that field — shifting from sequence-first to field-first design.

At ETH Zürich, as part of an XPRIZE finalist team, I am developing high-performance pipelines for biological applications on quantum computers — tackling strongly correlated metal centers intractable by classical methods. I apply electric field-guided and second coordination sphere engineering to heme and non-heme metalloenzymes, including cytochrome P450 variants for selective transformations and the catalytic mechanism of nitrogenase for sustainable nitrogen fixation. I also contribute to artificial metalloenzyme design for CO₂ capture and conversion, extending mechanistic insights to engineered catalysts for green chemistry. In drug discovery, I apply free energy perturbation simulations to characterize inhibitor binding to α-synuclein aggregation targets in Parkinson’s disease.