(369a) Computational protein design as a tool for insight, discovery, and surveillance

In this talk, we will discuss how molecular modeling and computational protein design can help inform diverse tasks ranging from improved enzymes, antibodies with higher affinities, protein pores tuned for targeted separations, and surveillance of emerging viral variants. At the heart of our algorithmic developments is the IPRO software (maranasgroup.com/software.htm) that uses biophysics-inspired scoring functions to quantify the binding energy between the redesigned protein and the target molecule(s). IPRO uses a mathematical programming core using mixed-integer linear (MILP) optimization to identify both the amino acid choices and rotamer conformations that bring about the desired binding characteristics. We will highlight recent predictions and experimental follow up studies aimed at (i) re-engineering of an acyl-ACP thioesterase with improved selectivity towards medium-chain fatty acids while maintaining overall high activity (with the Pfleger lab), (ii) the de novo design of anti-FLAG binding antibodies (with the Maynard lab), (iii) the redesign of protein pores embedded in a polymer membrane for solute separations (with the M. Kumar lab), and (iv) the computational surveillance of the potential for improved RBD-ACE2 binding affinity for human and livestock animals of current and potential future variants of SARS-CoV-2 (with the Kuchipudi lab). Concluding thoughts will focus on recent progress using machine learning tools in protein structure prediction, future perspectives and implications for protein design.