Research map / Probabilistic and scientific modeling / Molecules and proteins
Pallatom-Ligand: an All-Atom Diffusion Model for Designing Ligand-Binding Proteins
Haochen Wang, Qianyi Wang, Rui Ma, Jiawei Guan, Weikun.Wu, Haobo Wang, Jiayi Dou
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 |
|---|---|---|---|
| redesign sequences outside the interface with LigandMPNN | Pallatom-Ligand (w/out SA) · 8-ligand benchmark | Ca-RMSD: 2.80 ± 3.84 -> 1.39 ± 2.05 | Improved |
| redesign sequences outside the interface with LigandMPNN | Pallatom-Ligand (w/out SA) · 8-ligand benchmark | ipAE: 5.90 ± 2.94 -> 4.07 ± 1.98 | Improved |
| redesign sequences outside the interface with LigandMPNN | Pallatom-Ligand (w/ SA) · 8-ligand benchmark | ligand RMSD: 10.59 ± 7.47 -> 7.04 ± 5.37 | Improved |
| use Pallatom-Ligand (w/ SA, mpnn1) instead of RFdiffusion2 (mpnn1) | ligand-binding design · 8-ligand benchmark | Ca-RMSD: 3.94 ± 5.48 -> 1.36 ± 1.63 | Improved |
| use Pallatom-Ligand (w/ SA, mpnn1) instead of RFdiffusion2 (mpnn1) | ligand-binding design · 8-ligand benchmark | ligand RMSD: 7.12 ± 3.98 -> 7.04 ± 5.37 | No clear effect |
| use Pallatom-Ligand (w/out SA, mpnn1) instead of RFdiffusionAA (mpnn1) | ligand-binding design · 8-ligand benchmark | ipAE: 7.67 ± 3.28 -> 4.07 ± 1.98 | Improved |
| add ligand solvent-accessibility conditioning (all ligand atoms buried) (ablation) | Pallatom-Ligand · 8-ligand benchmark | protein-pLDDT: 84.48 ± 9.84 -> 81.87 ± 10.45 | Hurt |
| add alpha ratio condition 0.8-1.0 (ablation) | Pallatom-Ligand · 8-ligand benchmark, alpha-ratio control | protein-fold success rate: 71.5% -> 66.8% | Hurt |
| add alpha ratio condition 0.8-1.0 (ablation) | Pallatom-Ligand · 8-ligand benchmark, alpha-ratio control | ligand-pocket success rate: 4.4% -> 11.0% | Improved |
| add alpha ratio condition 0-0.2 (mainly beta) (ablation) | Pallatom-Ligand · 8-ligand benchmark, alpha-ratio control | protein-fold success rate: 71.5% -> 56.2% | Hurt |
| add alpha ratio condition 0.2-0.8 (ablation) | Pallatom-Ligand · 8-ligand benchmark, alpha-ratio control | diversity: 0.17 -> 0.26 | Improved |
| use Pallatom-Ligand w/out cond. instead of RFdiffusion2 | ligand-binding design · 8-ligand benchmark | ligand-pose success rate: 3.5% -> 1.0% | Hurt |
Closest papers by meaning
- P(all-atom) Is Unlocking New Path For Protein Design (ICML 2025 · similarity 0.90)
- Interaction-based Retrieval-augmented Diffusion Models for Protein-specific 3D Molecule Generation (ICML 2024 · similarity 0.86)
- Aligning Target-Aware Molecule Diffusion Models with Exact Energy Optimization (NeurIPS 2024 · similarity 0.86)
- An All-Atom Generative Model for Designing Protein Complexes (ICML 2025 · similarity 0.86)
- DecompOpt: Controllable and Decomposed Diffusion Models for Structure-based Molecular Optimization (ICLR 2024 · similarity 0.85)
- Harmonic Self-Conditioned Flow Matching for joint Multi-Ligand Docking and Binding Site Design (ICML 2024 · similarity 0.85)
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.
