Research map / Probabilistic and scientific modeling / Molecules and proteins
Static Recovery Is Not Dynamic Stability: Dynamics-Aware Benchmarking of Protein Motif Scaffolding
Emile de Bruyn, Moritz Schäffler, Jannik Schneider, Sebastian R Schmidt, Chanwoong Hwang, Aleena Siji, Utkarsh Upadhyay, Alexander Schug, Stefan Bauer, Artur Yakimovich, Stefan Kesselheim, Alina Bazarova
Read the paperCheck your idea against 137,217 papers
AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Argues that existing protein motif scaffolding benchmarks rely on small hand-curated motif sets and static refolding criteria, and proposes dynamics-aware benchmarking of scaffolds.
Problem
Protein motif scaffolding benchmarks use small hand-curated motif sets and static refolding-based success criteria (motif recovery, self-consistency, novelty), which may not capture whether scaffolds are dynamically stable.
Approach
Proposes a dynamics-aware benchmark for motif scaffolding that goes beyond static refolding criteria to assess dynamic stability of generated scaffolds, alongside broader motif sets (details not available in the abstract).
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use BILBY-PRO (220 cases) instead of MotifBench (30 cases) | model ranking · motif scaffolding leaderboard | rank of ESM3: 2nd (34.92) -> 5th (24.98) | Hurt |
| use BILBY-PRO instead of MotifBench | RFDiffusion · motif scaffolding | MotifBench score: 27.72 -> 62.20 | Improved |
| add MD-based dynamic checks on statically solved designs | static solved designs · BILBY-MD, 55 pairs | pass rate: 74.5% motif drift vs 63.6% scaffold drift vs 49.1% FES | Hurt |
| evaluate statically solved vs unsolved designs under MD criteria (ablation) | static solve label · BILBY-MD | MD-support (3 of 4 checks): 31/55 solved vs 6/79 unsolved | Improved |
| use token-based models (DPLM2, ESM3) instead of coordinate-based generators | scaffold stability · BILBY-MD | mean scaffold RMSD difference: 7.46 (FrameFlow) -> -0.54 (DPLM2) | Improved |
| compare compact local motifs vs interface motifs | motif type · BILBY-MD | MD support rate: 1.0 (06_6E6R) vs 0 (04_5WN9, 14_5IUS) | Mixed |
| rank models by coverage instead of MotifBench score | static evaluation · BILBY-PRO | model ranking | Mixed |
Surprising findings
- Static refolding-based criteria may conflate structural recovery with dynamic stability, per the title's framing.
Limitations
- Full text unavailable; details of the dynamics-aware criteria and motif sets are not verifiable from the abstract.
Experimental setup
Metrics: motif recovery self-consistency structural novelty
Closest papers by meaning
- ProteinBench: A Holistic Evaluation of Protein Foundation Models (ICLR 2025 · similarity 0.88)
- Inverse problems with experiment-guided AlphaFold (ICML 2025 · similarity 0.87)
- Simultaneous Modeling of Protein Conformation and Dynamics via Autoregression (NeurIPS 2025 · similarity 0.86)
- Ensembits: an alphabet of protein conformational ensembles (NeurIPS 2026 · similarity 0.86)
- TEMPO: Temporal Multi-scale Autoregressive Generation of Protein Conformational Ensembles (NeurIPS 2025 · similarity 0.86)
- SE(3)-Stochastic Flow Matching for Protein Backbone Generation (ICLR 2024 · similarity 0.86)
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.
