My favorite proteins are the enzymes which make possible (nearly) all of the nonspontaneous chemical reactions in the tree of life, inside and outside of the cell. Each one leverages multiple “strategies”, referred to by some scholars (including myself) as the “contributions to enzyme catalysis”. They weave together substrates using physical forces, hold them together until they react, convert thermal energy into directional motion, and they even use specialized tools (cofactors) for particularly difficult reactions. Pretty cool, right? I think so too!

Enzymes for global challenges

Since high school I’ve been learning how to solve and apply enzyme design towards difficult “global challenges” the world faces today: mitigating and reversing anthropogenic climate change, with special focus on carbon dioxide capture and pollutant biodegradation; enabling pharmaceutical synthesis at ambient temperature and pressure; and biosolar energy collection. I hope to build bioreactors for these reactions, and so promote a just transition to a renewable future: repairing the ecological impact of consumerist Western economies and fully supporting the material needs of peoples impacted by neo/colonialism.

Computational enzyme synthase design (CESD)

The enzymes we need to solve those problems haven’t evolved naturally yet, so we must design them ourselves. Nature shows us most of the clues though: not only can we improvise natural enzyme domains to accelerate elementary reaction rates, we can also learn from natural enzyme complexes and synthases to synthesize complex molecules or break down pollutants to simple metabolites. I’ll explain a more specific strategy in a blog post, soon.

I approach computational enzyme modeling using a few methods, especially: deep learning-based protein modeling and design, molecular dynamics simulation, and electronic structure methods such as DFT. I worked on RoseTTAFold-allatom and RoseTTAFold3 (RF3) in modelforge for three years in grad school, where I developed novel DL modules for automated refinement of protein structures into electron density maps. I’m trying to work other forms of publically-available biophysical data into these DL networks, such as: NMR chemical shifts, high-res XRD maps, and SAXS volumes.