Protein design rigorously tests our mastery of protein folding, stability and function. Protein design can be separated into redesign and de novo design by the issue of designability, which states that not all protein backbones will lead to viable sequences. The goal of redesign is to find favorable sequences for proteins with known structures, using their experimental coordinates as design models. De novo design requires design model coordinates to be created from scratch and then finds favorable sequences. Nature provides designable backbones in the case of fixed backbone redesign. In flexible redesign and de novo design, however, we have no guarantee of designability.

This work develops computational methods for flexible redesign and de novo design of diverse protein folds, probing questions of designability. We successfully used flexible redesign on several helix-bundle proteins and solved X-ray and NMR structures for one redesigned protein. The design model and the experimental structures are highly similar, < 1.0 Å backbone rmsd. Our success in de novo design has been modest. We have not succeeded in the de novo design of an all β-fold and continue to pursue this challenge. We have succeeded in the de novo design of a four helix-bundle protein. Preliminary NMR data suggests our design model and the experimental structure are the same fold and are similar at a global level with a backbone rmsd of ≤ 3Å.