GeoCycler: Reward-Aligned 3D Diffusion for Constraint-Conditioned Cyclic Peptide Design

2026-05-22

AI Summary

This work addresses the challenge that diffusion models face in de novo cyclic peptide design due to sparse, non-smooth, and highly combinatorial geometric closure constraints. The authors propose GeoCycler, a framework that employs a reward-weighted diffusion alignment mechanism to guide a conditional latent diffusion model during training toward learning closure-feasible cyclic peptide conformations. GeoCycler innovatively introduces a type-gated stepwise reward that provides dense geometric feedback conditioned on residue or linker types, combined with positive reward weighting and a replay buffer strategy to enable unified modeling of diverse cyclization topologies—including head-to-tail macrocycles and bicyclic structures. On the LNR benchmark, GeoCycler substantially improves the pass@5 closure success rate, outperforming CP-Composer by 20.8 percentage points for head-to-tail cyclization while preserving realistic amino acid compositions and backbone dihedral angle distributions.

Abstract

Cyclic peptides are attractive therapeutic modalities because their closed-ring topology can improve stability and target specificity. However, de novo cyclic peptide design remains challenging for diffusion generators, as macrocyclization requires satisfying sparse, non-smooth, and compositional geometric constraints. Existing constraint-conditioned methods largely rely on inference-time guidance, which can steer samples toward desired closures but does not directly change the learned generative distribution. We propose GeoCycler, a reward-weighted diffusion alignment framework for training conditional latent diffusion models toward macrocyclization feasibility. GeoCycler introduces a type-gated stair reward that activates distance-based shaping only when prerequisite residue or linker types are satisfied, providing dense geometric feedback while avoiding misleading signals from chemically incompatible anchors. Together with positive-only reward weighting and replay-based stabilization, GeoCycler aligns a single generator across multiple cyclization topologies. On the LNR benchmark, GeoCycler improves pass@5 closure success over strong guidance-based baselines across stapled, head-to-tail, disulfide, and bicyclic settings. In particular, it improves head-to-tail success by 20.8 percentage points over CP-Composer while maintaining comparable amino-acid and backbone-dihedral statistics. These results suggest that training-time alignment to sparse geometric constraints is a promising alternative to relying solely on post hoc sampling-time correction for cyclic peptide generation.

Problem

• cyclic peptide design
• macrocyclization
• geometric constraints
• diffusion generation
• constraint-conditioned generation

Innovation

• reward-weighted diffusion
• constraint-conditioned generation
• cyclic peptide design
• geometric alignment
• type-gated reward