Research map / Probabilistic and scientific modeling / Molecules and proteins
Learning residue level protein dynamics with multiscale Gaussians
Mihir Bafna, Bowen Jing, Bonnie Berger
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 log-Frobenius loss with MSE (ablation) | DynaProt-M · ATLAS marginals | RMSF r / sym KL: RMSF r 0.87 -> 0.38; Sym KL 0.91 -> 9.26 | Hurt |
| replace SE(3)-invariant IPA blocks with MLPs (ablation) | DynaProt-M · ATLAS marginals | RMSF r / RMWD: RMSF r 0.87 -> 0.48; RMWD 1.18 -> 1.92 | Hurt |
| explicit marginal prediction (DynaProt-M) vs FlexPert-3D with 1.2B params | DynaProt-M · ATLAS RMSF | RMSF r: 0.830 -> 0.865 | Improved |
| DynaProt-M vs NMA | DynaProt-M · ATLAS marginals | RMWD / Sym KL: 1.45 -> 1.18; 4.56 -> 0.91 | Improved |
| DynaProt-M vs AFMD+Templates | DynaProt-M · ATLAS marginals | RMWD median: 1.10 (AFMD+T) vs 1.18 | No clear effect |
| DynaProt-J vs NMA on residue coupling | DynaProt-J · ATLAS coupling | peak Pearson r: 0.59 -> 0.71 | Improved |
| DynaProt joint-reconstruction ensemble vs AFMD+T | DynaProt · ATLAS ensembles | global RMSF r / MD PCA W2: global RMSF r 0.91 -> 0.71; MD PCA W2 1.25 -> 1.74 | Mixed |
| DynaProt vs NMA ensembles | DynaProt · ATLAS ensembles | transient contacts J: 0.33 -> 0.29 | Hurt |
Closest papers by meaning
- Predicting directional flexibility in proteins (NeurIPS 2026 · similarity 0.90)
- Ensembits: an alphabet of protein conformational ensembles (NeurIPS 2026 · similarity 0.88)
- Simultaneous Modeling of Protein Conformation and Dynamics via Autoregression (NeurIPS 2025 · similarity 0.86)
- TEMPO: Temporal Multi-scale Autoregressive Generation of Protein Conformational Ensembles (NeurIPS 2025 · similarity 0.86)
- NMA-tune: Generating Highly Designable and Dynamics Aware Protein Backbones (ICML 2025 · similarity 0.86)
- Symmetry-Driven Discovery of Dynamical Variables in Molecular Simulations (ICML 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.
