Research map / Label-efficient and robust learning / Recognition and long-tail
Spherical convolutions and their application in molecular modelling
Wouter Boomsma, Jes Frellsen
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 |
|---|---|---|---|
| use concentric cubed-sphere convolution instead of Cartesian 3D convolution | CNN for protein environments · Amino acid prediction (Q20) | Q20: 0.500 -> 0.564 | Improved |
| use concentric cubed-sphere convolution instead of Cartesian 3D convolution | CNN for protein environments · Secondary structure (Q3) | Q3: 0.922 -> 0.933 | Improved |
| use cubed-sphere grid instead of spherical-polar grid | concentric spherical convolution · Amino acid prediction (Q20) | Q20: 0.560 -> 0.564 | No clear effect |
| use convolutions instead of dense layers (ablation) | CCS model · Amino acid prediction (Q20) | Q20: 0.348 -> 0.564 | Improved |
| use single conv over whole r range instead of conv-banded-disjoint (ablation) | CCS model · Amino acid prediction (Q20) | Q20: 0.515 -> 0.564 | Improved |
| use single conv over whole r range instead of conv-banded (ablation) | CCS model · Amino acid prediction (Q20) | Q20: 0.548 -> 0.564 | Improved |
| radial strategy variants (banded, banded-disjoint) vs single conv (ablation) | CCS model · Secondary structure (Q3) | Q3: 0.931 / 0.932 vs 0.933 | No clear effect |
| use concentric cubed-sphere instead of Cartesian features for ddG regression | ddG predictor · ddG stability, Kellogg | Pearson r: 0.66 vs 0.66 | No clear effect |
Closest papers by meaning
- Spherical Message Passing for 3D Molecular Graphs (ICLR 2022 · similarity 0.88)
- ATOM3D: Tasks on Molecules in Three Dimensions (NeurIPS 2021 · similarity 0.88)
- Spherical CNNs (ICLR 2018 · similarity 0.87)
- Equivariant Spherical Transformer for Efficient Molecular Modeling (NeurIPS 2026 · similarity 0.85)
- Beyond Atoms: Enhancing Molecular Pretrained Representations with 3D Space Modeling (ICML 2025 · similarity 0.85)
- Attention on the Sphere (NeurIPS 2025 · 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.
