The last few years have seen a large increase in the number of machine learning driven protein design tools. The Ellington lab uses these different tools in our protein engineering workflows to accelerate and improve the design of libraries. Starting from a defined structure or sequence, we use various sequence or structure-based models to redesign the protein for a defined property. This could include improved solubility or thermostability, enhanced enzyme activity, or improved binding affinity. Alternatively, there are protein design problems where you do not have a protein to use for redesign. Examples for this problem include designing a de novo binder to a small molecules or a protein protein target.

Associated Lab Members:

Clayton Kosonocky