
▲ AI-guided protein structure prediction
▲ AI-guided protein structure prediction
A protein language model using orders of magnitude less computing time than state-of-the-art AlphaFold2 while outperforming them on certain proteins. Enables rapid generation of designed proteins which fold, and screen them for function.
▲ Design protein pores
Protein pores offer naturally-optimized sieves for molecular and ionic separations. We learn how certains amino acid sequences fold into pores with selective interiors. We tune existing pores/ design them from scratch for a wide-range of separations.
▲ Protein-like synthetic channels
Nature-inspired synthetic channel geometries which can perform precise separations in aqueous phase have been identified. These channels are more robust to chemical and mechanical stresses than any natural channel. Applications include precise separation, energy-storing fuel cells.