We are creating a hybrid computational platform for studying protein–protein interactions and protein–ligand interactions that combines machine learning and physics-based approaches. Our approach relies on a geometric representation of the protein surface, which affords the opportunity to use advanced geometrical descriptors from the field of face recognition.
Antibody Selection for COVID-19
Spike proteins decorate the surface of the SARS-CoV-2 virus and play an important role in cell infection by binding to ACE2 receptors on the membrane of the human host cells. A possible strategy to prevent this bond from forming is to block the spike protein with synthetic antibodies. Our objective is to build a machine-learning model that can distinguish between antibodies that bind to spike protein and those that do not.