Flexibility-Aware Geometric Latent Diffusion for Full-Atom Peptide Design

Although peptides are well suited for flexible and shallow binding interfaces, their intrinsic flexibility induces a strongly coupled sequence–structure relationship that current fixed-geometry latent models cannot simultaneously model with conformational diversity and physical feasibility, ultimately limiting design quality. To overcome this bottleneck, PepFGLD is proposed as a receptor-conditioned, flexibility-aware framework for full-atom peptide design. The framework is motivated by a systematic analysis of existing limitations: geometry shifts driven by interfacial flexibility are not well captured by standard equivariant encoders; the static combination of sequence information and 3D geometry cannot represent their dynamic interactions; and diffusion models without timely geometric feedback tend to drift away from physically reasonable energy landscapes. In PepFGLD, FlexEGNN is used to improve the sensitivity of geometric representations to local flexibility, a coherent and adaptable latent conformational manifold is formed through bidirectional sequence–structure interaction and nonlinear latent mapping, and a time-dependent energy-guided diffusion mechanism is incorporated to balance exploration and convergence during diffusion so that sampling trajectories are continuously guided toward physically feasible full-atom structures. PepFGLD yields improved binding affinity and design success across multiple peptide design tasks.

Peptides are short chains of amino acids that can bind to target proteins and may support the development of new medicines. Computational peptide design remains difficult because peptides are highly flexible: they can bend, twist, and change structure as they interact with a protein. A useful design method therefore needs to identify not only a suitable amino acid sequence, but also a three-dimensional structure that is physically realistic. We introduce PepFGLD, a computational framework for designing peptides for a specified protein target. Rather than assuming that peptide structures are mostly fixed, PepFGLD focuses on flexible regions near the binding site and models how peptide sequence and structure affect each other during design. It also uses physical feedback during generation, helping guide the model toward structures whose atoms are arranged in more realistic ways. Across several peptide design tasks, PepFGLD generated candidates with improved predicted binding affinity and a higher rate of successful designs. These results suggest that accounting for peptide flexibility can make computer-aided peptide design more reliable and effective.