Proteins

Proteins & Geometric Deep Learning

Each protein not only has a unique sequence of amino acids — it also has a specific shape that determines its function. When protein folding goes awry, diseases like Alzheimer’s and Parkinson’s can occur. In recent years, researchers have made significant progress harnessing machine learning to predict protein folding, holding huge potential for drug development. However, current methods are limited in their ability to decipher the structure of particularly large or disordered proteins, or accurately predict protein binding. Our researchers are developing geometric deep learning methods to address this gap.