CODesign: Consistency from Data to Trajectory in All-Atom Protein Binder Co-Design

2026-10-01

AI Summary

This study addresses the challenge of cross-modal consistency in de novo protein design, where the mutual dependence between sequence and structure complicates generative modeling. To overcome this, the authors propose a consistency distillation approach for dataset construction alongside a consistency-aware joint resampling strategy. These techniques are integrated into a multimodal joint flow model that synergistically generates all-atom sequences, backbones, and side chains. The proposed framework enables the design of highly consistent binders, achieving state-of-the-art performance by improving the computational success rate by 70.9% on protein-ligand target binder tasks. Furthermore, all associated resources have been made publicly available to facilitate future research.

Abstract

The central challenge in de novo protein design is generating plausible, mutually compatible structures and sequences, such that each designed sequence folds into its intended structure and the structure accommodates that sequence. Compared to typical two-stage design methods, which decouple the modeling of the interdependent modalities, co-design models improve the cross-modal consistency by jointly generating sequences and structures. However, naively generating sequences and structures simultaneously does not ensure their consistency. To address this challenge, we propose CODesign framework. We improve data consistency by generating approximately 105,000 consistency-distilled dimers. We further promote consistency through a multimodal joint flow model that captures the joint distribution of sequences, backbone structures, and local atomic configurations, together with a consistency-aware joint resampling strategy that iteratively refines sequences and side chains. Experiments show that CODesign achieves state-of-the-art performance with the highest in silico success rates on both protein- and ligand-target binder design. Ablation studies also demonstrate our distilled dataset increases performance by 70.9%, which can be further improved by our proposed resampling mechanism with negligible additional computational cost. Code, model weights and the new dataset will be completely open-source.

Problem

• de novo protein design
• protein binder co-design
• sequence-structure consistency
• all-atom generation

Innovation

• Protein Binder Co-Design
• Consistency Distillation
• Multimodal Joint Flow Model
• Joint Resampling Strategy
• All-Atom Design