Research map / Probabilistic and scientific modeling / Molecules and proteins
Iterative Refinement Graph Neural Network for Antibody Sequence-Structure Co-design
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Use iterative sequence-structure co-design instead of sequence-only generation | RefineGNN · SAbDab CDR-H3 test | PPL: 9.70 -> 8.38 | Helped |
| Use iterative global structure refinement rather than AR-GNN generation | CDR-H3 generation · SAbDab CDR-H3 test | RMSD: 3.63 -> 2.50 | Helped |
| Add context attention (ablation) | RefineGNN · SAbDab CDR-H3 test | PPL: 8.86 -> 8.38 | Helped |
| Add context coarsening (ablation) | RefineGNN · SAbDab CDR-H3 test | PPL: 9.01 -> 8.38 | Helped |
| Remove both attention and context coarsening, making generation unconditional (ablation) | RefineGNN · SAbDab CDR-H3 test | PPL | Hurt |
| Halve the number of refinement steps (ablation) | RefineGNN · SAbDab CDR-H3 test | PPL | Hurt |
| Increase context block size to b=8 (ablation) | RefineGNN · SAbDab CDR-H3 test | PPL | Hurt |
| Feed the ground-truth structure at every generation step (ablation) | RefineGNN · SAbDab CDR-H3 test | PPL: 8.38 -> 7.39 | Helped |
| Use RefineGNN for property-guided antibody optimization instead of the best baseline | SARS-CoV-2 neutralization optimization · CoVAbDab test set | neutralization score: 72.0% -> 75.2% | Helped |
| Use RefineGNN pretrained on SAbDab instead of AR-GNN | CDR-H3 generation · CoVAbDab | PPL: 8.67 -> 7.86 | Helped |
Closest papers by meaning
- Cross-Gate MLP with Protein Complex Invariant Embedding Is a One-Shot Antibody Designer (AAAI 2024 · similarity 0.90)
- Relation-Aware Equivariant Graph Networks for Epitope-Unknown Antibody Design and Specificity Optimization (AAAI 2025 · similarity 0.89)
- Antigen-Specific Antibody Design via Direct Energy-based Preference Optimization (NeurIPS 2024 · similarity 0.88)
- Antibody Design Using a Score-based Diffusion Model Guided by Evolutionary, Physical and Geometric Constraints (ICML 2024 · similarity 0.88)
- Pareto-Optimal Energy Alignment for Designing Nature-Like Antibodies (NeurIPS 2025 · similarity 0.87)
- Retrieval Augmented Diffusion Model for Structure-informed Antibody Design and Optimization (ICLR 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.