RNADyn: A Benchmark for Generating and Understanding RNA Dynamics
RNADynBench ships 2,585 quality-controlled 100-ns RNA molecular dynamics trajectories, plus RNADynNet, a unified model for trajectory generation and dynamics fingerprinting.
The paper introduces RNADynBench, a standardized RNA molecular dynamics benchmark containing 2,585 quality-controlled 100-ns all-atom trajectories with leakage-controlled splits. On top of it, RNADynNet combines coordinate denoising, single-frame-to-trajectory alignment, and physical grounding in a shared backbone for both trajectory generation and dynamics representation learning. Generated trajectories reach RMSF correlations of 0.875 and 0.766 across two test sets, including a high-flexibility challenge set, and single-conformer predictions agree comparably with MD-derived dynamics.
- Benchmark spans 2,585 all-atom 100-ns RNA trajectories with leakage-controlled splits
- Single shared-backbone model handles both trajectory generation and dynamics fingerprint extraction
- Physical grounding improves generated dynamics and recoverable physical information
- RMSF correlations of 0.875 and 0.766 achieved on the two test sets
- Organizations
- RNADynBenchRNADynNet
From the original article
1 min read at arxiv.orgRibonucleic acid (RNA) functions through conformational changes that are not fully captured by static structures. However, large-scale standardized RNA dynamics data remain limited, and existing approaches typically treat trajectory generation and dynamics understanding as separate objectives.
By Yiming Huang · arxiv.org
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.