Interpretable antibody-antigen interaction prediction by introducing route and priors guidance

With the application of personalized and precision medicine, more precise and efficient antibody drug development technology is urgently needed. Identification of antibody-antigen interactions is crucial to antibody engineering. The time-consuming and expensive nature of wet-lab experiments calls for efficient computational methods. Taking into account the non-overlapping advantage of current structure-dependent and sequence-only computational methods, we propose an interpretable antibody-antigen interaction prediction method, S3AI. The introduction of structural knowledge, combined with explicit modeling of chemical rules, establishes a 'sequence-to-function' route in S3AI, thereby facilitating its perception of intricate molecular interactions through providing route and priors guidance. S3AI significantly and comprehensively outperforms the state-of-the-art models and exhibits excellent generalization when predicting unknown antibody-antigen pairs, surpassing specialized prediction methods designed for out-of-distribution generalization in fair comparisons. More importantly, S3AI captures the universal pattern of antibody-antigen interactions, which not only identifies the CDRs responsible for specific binding to the antigen but also unearths the importance of CDR-H3 for the interaction. Structure-free design and superior performance make S3AI ideal for large-scale, parallelized antibody optimization and screening, enabling the rapid and precise identification of promising candidates within the extensive antibody space. ### Competing Interest Statement The authors have declared no competing interest.