ABOPD: Antibody CDR Design via On-Policy Distillation

2026-07-21

AI Summary

Existing protein generation models suffer from inconsistent training and inference trajectories in antibody CDR-H3 loop design, leading to accumulated backbone bias and geometric distortion. This work proposes ABOPD, a novel framework that introduces in-policy distillation into antibody design for the first time. By leveraging privileged geometric information from native structures during the denoising diffusion process, ABOPD provides fine-grained supervision over intermediate generated states, enabling precise conformational correction at each step. This approach effectively bridges the distributional gap between training and inference, significantly improving structural fidelity. On the RAbD benchmark, ABOPD reduces the CDR-H3 structure recovery RMSD by 0.42 Å—from 2.37 Å to 1.95 Å—outperforming baseline methods such as supervised fine-tuning and offline distillation.

Abstract

Antibodies are essential therapeutic molecules, and their complementarity-determining regions (CDRs) form the primary antigen-recognition interface. Recent protein generative models have demonstrated broad capabilities in biomolecular design, yet post-training strategies for downstream objectives remain limited. Standard denoising training operates on noisy states obtained by perturbing native structures, whereas recursive generation proceeds through model-generated intermediate states. For flexible antibody CDR loops such as CDR-H3, this mismatch can allow backbone deviations to accumulate along the denoising trajectory and compromise antigen-facing loop geometry. We introduce ABOPD, an antibody design framework based on on-policy distillation that leverages privileged native geometry during training to supervise states visited along the model's own denoising trajectories. With this fine-grained structural supervision, ABOPD substantially improves structural recovery on RAbD CDR-H3 generation, reducing RMSD by 0.42 Å (from 2.37 Å to 1.95 Å) and outperforming supervised fine-tuning and offline distillation controls, offering a path to higher-fidelity protein design.

Problem

• antibody CDR design
• CDR-H3 loop
• structural fidelity
• denoising trajectory
• backbone deviation

Innovation

• on-policy distillation
• antibody CDR design
• protein generative modeling
• structural recovery
• CDR-H3 loop