Computationally predicting peptide-MHC class II (pMHCII) interactions is critical for vaccine design and autoimmune disease research, yet remains challenging due to MHCII polymorphism and structural complexity. We propose GeoMHCII, a geometric deep learning framework that synergizes evolutionary information with structural modeling to achieve pan-allele generalization. By leveraging E(n)-equivariant graph networks on co-folded pMHCII structures and integrating pretrained protein language embeddings, the model effectively captures both spatial and evolutionary determinants of binding specificity. Systematic benchmarking against experimental structures validates the accuracy of co-folding models, underscoring their utility for binding prediction tasks. Extensive experiments demonstrate that the proposed model outperforms existing methods, achieving a 5.2 % gain in AUCROC on a balanced test split compared to the strongest baseline. Notably, GeoMHCII maintains robust performance in high-confidence structure cases where the distinction between binders and non-binders is subtle, and its output can further refine the structural pool to reflect more realistic binding patterns. This work advances structureaware immunoinformatics, providing a robust tool for epitope discovery and therapeutic development. The code is available at: https://github.com/YaokunJi/GeoMHCII

Antibodies are essential proteins responsible for immune responses in organisms, capable of specifically recognizing antigen molecules of pathogens. Recent advances in generative models have significantly enhanced rational antibody design. However, existing methods mainly create antibodies from scratch without template constraints, leading to model optimization challenges and unnatural sequences. To address these issues, we propose a retrieval-augmented diffusion framework, termed RADAb, for efficient antibody design. Our method leverages a set of structural homologous motifs that align with query structural constraints to guide the generative model in inversely optimizing antibodies according to desired design criteria. Specifically, we introduce a structure-informed retrieval mechanism that integrates these exemplar motifs with the input backbone through a novel dual-branch denoising module, utilizing both structural and evolutionary information. Additionally, we develop a conditional diffusion model that iteratively refines the optimization process by incorporating both global context and local evolutionary conditions. Our approach is agnostic to the choice of generative models. Empirical experiments demonstrate that our method achieves state-of-the-art performance in multiple antibody inverse folding and optimization tasks, offering a new perspective on biomolecular generative models.