MultiSAAl: Sequence-Informed Antibody-Antigen Interaction Prediction Using Multiscale Deep Learning

Antibody-antigen interaction prediction is essential for therapeutic development but remains experimentally costly. The dynamic conformational changes essential to antibody-antigen binding are often missed by structure-based methods relying on static snapshots, underscoring the need for accurate sequence-based approaches. We propose MultiSAAI, a sequence-informed framework that models antibody-antigen interactions by explicitly accounting for the distinct roles of antibody heavy and light chains in antigen binding. MultiSAAI integrates language model embeddings, physicochemical properties, geometric constraints, and residue substitutability to characterize antibody-antigen interactions across multiple scales, employing a multiscale network architecture that simultaneously evaluates global residue-pair compatibility and local amino acid fitness at the binding interface. Furthermore, the incorporation of site-specific information and biologically grounded binding principles allows the model to more closely reflect the actual mechanisms of interactions. Benchmark results demonstrate that MultiSAAI achieves AUROC scores of 0.772 on the generic antibody-antigen interaction data set and 0.947 on the SARS-CoV-2 data set, outperforming existing methods such as A2binder and AbAgIntPre. Finally, large-scale preliminary antibody screening further validates the potential of MultiSAAI for high-throughput therapeutic antibody discovery.