Celebrating the 2024 Nobel Prize in Chemistry. Computational prediction: An authentic revolution. Methods for the structural prediction and design of proteins

Abstract

About 65 years ago, X-ray crystallography revealed the first protein tertiary structures, showing the link between the folding and function of such structures. Since Anfinsen’s 1972 Nobel Prize, a major challenge of structural biology has been to predict a structure from its sequence — and, conversely, to design sequences that adopt a chosen shape. The 2024 Nobel Prize in Chemistry honored David Baker for his methods of computational protein design and Demis Hassabis and John Jumper for Alpha- Fold2, a neural network able to predict the folded structure of proteins with high accuracy, in some cases approaching experimental resolution. Although AF2 was built mainly to predict a single structure, researchers have found that by tweaking its initial parameters AF2 can output multiple structures — that is, several conformations for one same amino-acid sequence. These AI-based tools have boosted the application of artificial intelligence-based strategies for computational protein design. Despite the advances, though, one of today’s major challenges is the accurate computational prediction of proteins with a specific high catalytic activity.