Research

Brief summary

De novo protein design has experienced a machine learning-fueled revolution, vastly expanding our ability to design complex and functional proteins beyond those found in nature. Our research group works at the intersection of de novo protein design, deep learning and fundamental biophysics of protein function, doing both computational design and wet lab biophysical characterization. 
We aim to deconstruct biological function by reconstructing such function de novo. Using deep learning-based protein design, we construct proteins free from evolutionary constraints that incorporate desired functions to better understand their underlying biological mechanism. Among other areas, we are interested in mechanisms at the host-pathogen interface: 

• Creating synthetic “harpoons” that can covalently target selected epitopes by reconstructing an autocatalytic mechanism from gram-positive pathogens, a potential tool for irreversible, covalent opsonization. 
• Designing self-strengthening catch-bonds, counterintuitive protein-protein interactions that bind more tightly under mechanical stress. These de novo catch bonds will be minimal models to build an understanding of their mechanism and could become the basis for novel biomaterials. 

There are more potential projects that we will pursue, among them work regarding stabilizing natural proteins with desired functions such as enzymes and designing protein binders to therapeutically or fundamentally relevant native proteins.
We leverage our recently developed medium/high throughput screening technology to test designs (96-192 proteins per day), including biophysical characterization and Atomic Force Microscopy-based single-molecule protein folding studies. Thus, we can rapidly take an idea, transform it into a protein blueprint on the computer and finally validate our designs it in the wet lab.