2/16 Laufer Center Seminar

High-throughput discovery of protein folding stability and dynamics.

Protein sequences vary widely in folding stability, and optimizing stability remains a key challenge in protein engineering. Protein sequences also vary in their conformational dynamics: some sequences are "locked" into one rigid structure while others sample a landscape of diverse folded and partially-folded conformations. Machine learning methods promise to empower protein engineers to precisely tune stability and dynamics for specific applications, ushering in a new era of protein design. However, developing accurate tools is bottlenecked by the limited availability of experimental data measuring protein stability and dynamics. Our lab has pioneered high-throughput methods to quantify stability and dynamics on unprecedented scales, including recently released stability measurements for >750,000 protein domains and unpublished measurements quantifying conformational dynamics for >5,000 domains. I will share our lab's approach to developing these high-throughput assays as well as our findings on how protein sequences determine stability and dynamics. Furthermore, by comparing our stability measurements to our independent dynamics measurements, I will illustrate the complementarity between these approaches and the unique insights gained by pursuing both. Finally, I will share progress in developing predictive machine learning tools based on these new high-throughput data.