MuCO: Generative Peptide Cyclization Empowered by Multi-stage Conformation Optimization

2026-01-30

arXiv.org

AI Summary

Cyclic peptides exhibit rich conformational diversity that cannot be accurately captured by linear peptide folding models, hindering their virtual screening as drug candidates. To address this challenge, this work proposes MuCO, a novel method that decouples conformation generation into three stages: topology-aware backbone design, generative side-chain placement, and physics-informed all-atom refinement. MuCO introduces a pioneering multi-stage parallel sampling framework that efficiently explores diverse low-energy conformations while preserving physical plausibility. Extensive experiments on the large-scale CPSea dataset demonstrate that MuCO significantly outperforms current state-of-the-art methods in structural diversity, physical stability, secondary structure recovery, and computational efficiency.

Abstract

Modeling peptide cyclization is critical for the virtual screening of candidate peptides with desirable physical and pharmaceutical properties. This task is challenging because a cyclic peptide often exhibits diverse, ring-shaped conformations, which cannot be well captured by deterministic prediction models derived from linear peptide folding. In this study, we propose MuCO (Multi-stage Conformation Optimization), a generative peptide cyclization method that models the distribution of cyclic peptide conformations conditioned on the corresponding linear peptide. In principle, MuCO decouples the peptide cyclization task into three stages: topology-aware backbone design, generative side-chain packing, and physics-aware all-atom optimization, thereby generating and optimizing conformations of cyclic peptides in a coarse-to-fine manner. This multi-stage framework enables an efficient parallel sampling strategy for conformation generation and allows for rapid exploration of diverse, low-energy conformations. Experiments on the large-scale CPSea dataset demonstrate that MuCO significantly outperforms state-of-the-art methods in consistently in physical stability, structural diversity, secondary structure recovery, and computational efficiency, making it a promising computational tool for exploring and designing cyclic peptides.

Problem

• peptide cyclization
• conformation modeling
• cyclic peptides
• virtual screening
• structural diversity

Innovation

• peptide cyclization
• conformation generation
• multi-stage optimization
• generative modeling
• cyclic peptides