Reproduce “THFlow: A Temporally Hierarchical Flow Matching Framework for 3D Peptide Design” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Dengdeng Huang, Shikui Tu
THFlow, a new Temporally Hierarchical Flow Matching framework from Shanghai Jiao Tong University, addresses multimodal temporal inconsistency in 3D peptide design by decoupling the evolution of peptide position from other modalities. It generates peptides with superior binding affinity and stability, achieving 17.81% affinity on co-design and the lowest RMSD of 4.09 for peptide re-docking.
Key insights
- Explicitly modeling the temporal hierarchy of different peptide modalities during docking significantly improves the biological realism and efficacy of generative design.
- Integrating physicochemical interaction features into the model's representation is critical for designing peptides with superior binding affinity and stability.
- Dynamically balancing loss terms for structural and sequence alignment during training is essential for coherent co-design of peptide sequence and 3D structure.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.