Evolution guided atomistic design - a new paradigm for design of function

Our lab's long-term goal is to enable reliable and completely computational design of efficient, selective, and stable protein binders and enzymes. To achieve this goal, we are developing a unique strategy called evolution-guided atomistic design that uses information encoded in the evolutionary history of protein families to infer what structure and sequence features are likely to be tolerated in any given protein. We then use these rules to guide Rosetta atomistic design calculations in the search for new proteins with desired functions. To test our algorithms, we design new proteins that don't exist in nature and carry out wet-lab experiments either in-house or with our collaborators. Feedback from these experiments then enables the development of more sophisticated design algorithms. We therefore combine cutting-edge computational methods development with high-throughput experimental screening and stringent biochemical and structural analysis of designed proteins.

Automated protein optimisation

Proteins in nature are subject to evolutionary processes that optimise their activities. In recent decades, scientists have emulated natural evolutionary process in the lab, enabling the optimisation of proteins for human needs, thus generating efficient enzymes and binders. But evolution is an iterative process in which every change in a protein (mutation) must result in a variant that is at least as functional as its predecessor or it would be purged by the powerful forces of selection. Thus, lab evolution experiments may take years of tedious trial-and-error. We developed several methods that enable rapid, one-shot optimisation of protein activities, generating enzymes that degrade a broad spectrum of highly toxic nerve agents, antibodies with much improved affinity and stability, and even a much cheaper and more stable variant of a protein that is the prime candidate to serve as a vaccine for malaria. One of the most important goals for the lab is to enable broad use of our algorithms by biochemists and protein engineers, and we therefore develop web servers that allow researchers around the world to customise our design protocols for their particular needs. The web servers carry out the calculations on our lab's computer cluster and return models of improved binders and enzymes by email. You're most welcome to try these web servers yourself!

Designing new protein structures

Control over protein activity demands control over the protein backbone structure, but the backbone has numerous degrees of freedom and design of new backbones in protein active sites has been a notoriously difficult problem. By combining information from naturally occurring structures and sequences with atomistic design calculations, we developed a new approach for backbone design in active sites. Initially, we implemented this strategy to design new antibodies, and therefore called the method AbDesign. Encouraged by the method's success in designing atomically accurate new antibodies with over 50 mutations from any naturally occurring antibody, we next applied this method to the design of high-efficiency new enzymes and a large network of interacting pairs of proteins that exhibited ultrahigh specificity binding. Thus, evolution-guided atomistic design provides exquisite control over protein structure, stability, and activity.

Where are we headed?

The holy grail of our field is to enable the complete computational design of any arbitrarily chosen biomolecular activity. To enable such template-free design of function, we still need to learn a lot more about how function is encoded in proteins. We are therefore developing a new approach to design not a handful of binders or enzymes as in all current protein design methods, but vast repertoires comprising millions of substantially different variants. We use high-throughput screening methods to isolate the functional designs and deep sequencing analysis to fully characterise their activity. Next, advanced machine-learning methods are trained to find molecular features that discriminate the best designs from the rest, and these features are then used to improve the design algorithms, leading to a continuous, unbiased and systematic approach to learn the rules for designing new biomolecular activities. We are applying this strategy to the design of new hydrolytic enzymes, single-domain camelid antibodies, and small-molecule binders.