Research map / Probabilistic and scientific modeling / Molecules and proteins
Multi-Scale Representation Learning on Proteins
Vignesh Ram Somnath, Charlotte Bunne, Andreas Krause
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 |
|---|---|---|---|
| add surface layer to structure layer (multi-scale) (ablation) | HoloProt · PDBbind identity 30% | RMSE: 1.476 -> 1.464 | Improved |
| add structure layer to surface layer (ablation) | HoloProt · PDBbind identity 30% | RMSE: 1.482 -> 1.464 | Improved |
| add surface layer to structure layer (ablation) | HoloProt · Enzyme reaction class | accuracy: 74.2 -> 77.8 | Improved |
| add structure to surface-only model (ablation) | HoloProt · Enzyme reaction class | accuracy: 28.6 -> 77.8 | Improved |
| add surface layer to structure layer (ablation) | HoloProt · PDBbind identity 60% | RMSE: 1.378 -> 1.365 | Improved |
| add surface layer to structure layer (ablation) | HoloProt · PDBbind scaffold | RMSE: 1.521 -> 1.523 | No clear effect |
| replace full surface with molecular superpixels | HoloProt · PDBbind identity 30% | RMSE: 1.464 -> 1.491 | No clear effect |
| replace full surface with molecular superpixels | HoloProt · Enzyme reaction class | accuracy: 77.8 -> 78.9 | Improved |
| replace full surface with molecular superpixels | HoloProt · PDBbind scaffold | RMSE: 1.523 -> 1.516 | Improved |
| learn superpixel patch features with MPN instead of summary statistics (ablation) | HoloProt superpixels · PDBbind identity 30% | RMSE: 1.491 -> 1.491 | No clear effect |
| learn superpixel patch features with MPN instead of summary statistics (ablation) | HoloProt superpixels · Enzyme reaction class | accuracy: 78.9 -> 75.7 | Hurt |
Closest papers by meaning
- AtomSurf: Surface Representation for Learning on Protein Structures (ICLR 2025 · similarity 0.88)
- Towards Multiscale Graph-based Protein Learning with Geometric Secondary Structural Motifs (NeurIPS 2025 · similarity 0.88)
- Evaluating Representation Learning on the Protein Structure Universe (ICLR 2024 · similarity 0.88)
- Multi-Scale Representation Learning for Protein Fitness Prediction (NeurIPS 2024 · similarity 0.86)
- Learning Complete Protein Representation by Dynamically Coupling of Sequence and Structure (NeurIPS 2024 · similarity 0.85)
- Clustering for Protein Representation Learning (CVPR 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.
