Computational protein design has become a powerful approach for creating valuable proteins. By creating proteins that do not exist in nature, unmet biomedical or biotechnological needs can be met. We have developed a computational pipeline to discover antibodies that bind to a pre-defined surface of target proteins. Our approach is showcased by discovering and developing a human neutralizing antibody that binds to the receptor-binding domain (RBD) of the SARS-CoV-2 spike glycoprotein of all currently circulating variants of the virus, including Omicron, with potent affinity (pico- to femtomolar dissociation constants). We do physics-based calculations (using ROSETTA software suite) and learning-based computations (deep learning or machine learning). We focus on the design of therapeutic proteins (including monoclonal antibodies) for cancer immunotherapy and neuropathological diseases.
