Engineering enzymes for enhanced specific activity

How do enzymes function, and how can we make them more efficient?

With Professor Bruce Tidor, I have developed computational methods for rationally redesigning an enzyme for increased activity on a native substrate. Currently, most rational, structural approaches primarily focus on the energetic gap between the enzyme-bound ground state and transition state, without explicit consideration of the enzyme-substrate complex's dynamical approach toward, and along, the energy surface connecting these two important states. The methods I've developed capture and use this information, relate it to catalytic efficiency, and apply it to assist in enzyme redesign for enhanced activity.

Different aspects of this work have been presented at the national meetings of the Biophysical Society, the American Chemical Society, and the Protein Society, and we will soon submit our results for publication.

Using deep learning to map the relationship between catalytic activity and enzyme-substrate dynamics

Can we predict catalytic activity from only pre-reaction dynamics?

I have developed deep learning models that predict an enzyme mutant's level of catalytic activity based on its pre-reaction dynamics. That is, these models can predict whether a mutant is slower or faster than WT based on how it behaves within a few hundred femptoseconds before attempting to react. You can find a demonstration of these models on my GitHub.