To robustly design new proteins and predict their stability and interactions with other proteins, we must understand the geometric and physico-chemical principles underlying protein structure. Despite the abundance of structural data, we do not have a predictive, quantitative understanding of protein backbone and side-chain conformations. Our approach is to apply a simple physical model that includes only a minimal set of geometrical constraints, excluded volume, and attractive van der Waals interactions to:
- Predict the observed side-chain dihedral angle distributions of all hydrophobic and non-charged amino acids.
- Predict the local structure and rank the binding affinity of designed protein-protein interactions and predict the local structure and rank the stability of proteins with ‘repacked’ hydrophobic cores.
This work will not only lead to an enhanced fundamental understanding of protein structure and stability but will also enable us to develop efficient computational methods to rationally design protein interfaces with tunable specificity and affinity. Such designer proteins have numerous applications in biotechnology and medicine.
In the video below, Diego talks about the research he has conducted in the O’Hern lab, in collaboration with the Regan lab. This research is published at D. Caballero, J. Määttä, A. Q. Zhou, M. Sammalkorpi, C. S. O’Hern, and L. Regan, “The intrinsic α-helical and β-sheet preferences: A computational case study of Alanine”, Protein Science 23 (2014) 970.