antibody

SurfBind, a surface-centric learning framework for epitope prediction, uses Transformer-based architecture with patch-level surface modeling and binder-aware cross-attention to achieve state-of-the-art performance on epitope identification benchmarks.