Scalable design of repeat protein structural dynamics via probabilistic coarse-grained models

Computational protein design has emerged as a powerful tool for creating proteins with novel functionalities. However, most existing methods ignore structural dynamics even though they are known to play a central role in many protein functions. Furthermore, methods like molecular dynamics that are able to simulate protein movements are computationally demanding and do not scale for the design of even moderately sized proteins. Here, we develop a probabilistic coarse-grained model to overcome these limitations and support the design of the structural dynamics of modular repeat proteins. Our model allows us to rapidly calculate the probability distribution of structural conformations of large modular proteins, enabling efficient screening of design candidates based on features of their dynamics. We demonstrate this capability by exploring the design landscape of 4-6 module repeat proteins. We assess the flexibility, curvature and multi-state potential of over 65,000 protein variants and identify the roles that particular modules play in controlling these features. Although our focus here is on protein design, the methods developed are easily generalised to any modular structure (e.g., DNA origami), offering a means to incorporate dynamics into diverse biological design workflows.