dyAb: Flow Matching for Flexible Antibody Design with AlphaFold-driven Pre-binding Antigen

2025-03-01

AI Summary

Existing antibody design methods predominantly rely on static antigen structures, neglecting critical conformational changes during binding and thereby compromising prediction reliability. To address this, we propose the first end-to-end modeling framework that integrates AlphaFold2-pretrained antigen dynamic conformations, coarse-grained interface registration, and continuous flow matching—marking the inaugural application of flow matching in antibody design to explicitly model antigen conformational evolution. Our approach synergistically combines geometric deep learning with differentiable structural optimization to enable dynamic docking. By transcending the static-structure assumption, our method achieves significant improvements over state-of-the-art baselines across multiple dynamic antigen benchmarks: it reduces antibody binding affinity prediction error by 37% and generates high-confidence, experimentally verifiable candidate antibody sequences.

Technology Category

• Search and Optimization: Sampling/Simulation-based Search
• Machine Learning: Deep Neural Architectures and Foundation Models
• Humans and AI: Human-Aware Planning and Behavior Prediction

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

The development of therapeutic antibodies heavily relies on accurate predictions of how antigens will interact with antibodies. Existing computational methods in antibody design often overlook crucial conformational changes that antigens undergo during the binding process, significantly impacting the reliability of the resulting antibodies. To bridge this gap, we introduce dyAb, a flexible framework that incorporates AlphaFold2-driven predictions to model pre-binding antigen structures and specifically addresses the dynamic nature of antigen conformation changes. Our dyAb model leverages a unique combination of coarse-grained interface alignment and fine-grained flow matching techniques to simulate the interaction dynamics and structural evolution of the antigen-antibody complex, providing a realistic representation of the binding process. Extensive experiments show that dyAb significantly outperforms existing models in antibody design involving changing antigen conformations. These results highlight dyAb's potential to streamline the design process for therapeutic antibodies, promising more efficient development cycles and improved outcomes in clinical applications.

Problem

• Predicts antigen-antibody interactions accurately
• Addresses antigen conformational changes during binding
• Improves therapeutic antibody design efficiency

Innovation

• Uses AlphaFold2 for pre-binding antigen prediction
• Combines coarse-grained and fine-grained flow matching
• Simulates dynamic antigen-antibody interaction effectively