Research map / Probabilistic and scientific modeling / Molecules and proteins
Full-Atom Peptide Design via Riemannian–Euclidean Bayesian Flow Networks
Hao Qian, Shikui Tu, Lei Xu
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| replace a single Gaussian with a Gaussian mixture for side-chain torsion Bayesian flow (ablation) | PepBFN side-chain packing · Side-chain packing | correct % (within 20 deg): 59.54 -> 75.24 | Improved |
| replace a single Gaussian with a Gaussian mixture (ablation) | PepBFN side-chain packing · Side-chain packing | chi1 MAE (deg): 24.50 -> 10.75 | Improved |
| replace a single Gaussian with a Gaussian mixture (ablation) | PepBFN side-chain packing · Side-chain packing | chi2 MAE (deg): 32.44 -> 12.26 | Improved |
| replace a single Gaussian with a Gaussian mixture (ablation) | PepBFN side-chain packing · Side-chain packing | chi3 MAE (deg): 56.22 -> 34.25 | Improved |
| replace a single Gaussian with a Gaussian mixture (ablation) | PepBFN side-chain packing · Side-chain packing | chi4 MAE (deg): 60.35 -> 53.21 | Improved |
| model peptides in fully continuous Bayesian parameter space instead of hybrid continuous-discrete space (ablation) | PepFlow · Peptide binder design | convergence speed and sequence stability | Improved |
| use PepBFN_sc instead of the best baseline DiffPack | DiffPack · Side-chain packing | correct % (within 20 deg): 62.58 -> 75.24 | Improved |
| use PepBFN_seq instead of PepFlow for reverse folding | PepFlow · Peptide reverse folding | AAR %: 53.98 -> 62.46 | Improved |
| use PepBFN_seq instead of PepFlow for reverse folding | PepFlow · Peptide reverse folding | Hamming diversity: 26.40 -> 13.21 | Hurt |
| use PepBFN instead of PepFlow for co-design | PepFlow · Peptide sequence-structure co-design | binding affinity %: 21.37 -> 22.26 | Improved |
| use PepBFN instead of PepFlow for co-design | PepFlow · Peptide sequence-structure co-design | RMSD (A): 2.07 -> 3.58 | Hurt |
| use PepBFN instead of RFDiffusion for co-design | RFDiffusion · Peptide sequence-structure co-design | stability %: 26.82 -> 17.56 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- DiffPack — DiffPack: A Torsional Diffusion Model for Autoregressive Protein Side-Chain Packing (NeurIPS 2023) · better on 2 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Full-Atom Peptide Design based on Multi-modal Flow Matching (ICML 2024 · similarity 0.92)
- Full-Atom Peptide Design with Geometric Latent Diffusion (NeurIPS 2024 · similarity 0.90)
- PPFLOW: Target-Aware Peptide Design with Torsional Flow Matching (ICML 2024 · similarity 0.89)
- SE(3)-Stochastic Flow Matching for Protein Backbone Generation (ICLR 2024 · similarity 0.87)
- AlphaFold Meets Flow Matching for Generating Protein Ensembles (ICML 2024 · similarity 0.87)
- PepTune: De Novo Generation of Therapeutic Peptides with Multi-Objective-Guided Discrete Diffusion (ICML 2025 · similarity 0.87)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
