Andreea Deac

This paper predicts antibody-antigen binding sites using geometric deep learning methods, finding that surface-based models outperform others and achieving state-of-the-art results.