Dyna-1: Predicting Millisecond Protein Dynamics from Missing NMR Signals in Nature
sokrypton · x · 2026-08-11
A research team published the final version of Dyna-1, an AI model for predicting protein dynamics, in Nature, alongside fully open-sourced datasets.
- Background: Protein functions rely heavily on microsecond to millisecond conformational interconversions, but a lack of standardized, large-scale experimental data has hindered predictive modeling.
- Methodology: After curating over 100 NMR relaxation datasets, the team realized millisecond dynamics cause NMR signals to broaden beyond detection, leaving missing residues in chemical shift data. Dyna-1 learns from these "missing" signals.
- Key Updates: The final manuscript includes prospective experimental validation, confirming the model's real-world predictive power.
Related event: Nature Publishes Protein Dynamics Model Dyna-1(2 posts)→
More from Research
- Training on experimental function improves prediction of tumor-reactive T-cell receptors — bravo_abad · 2026-09-21
- MBZUAI builds value-conflict dataset and task vectors to steer LLM ethics — MBZUAI · 2026-09-21
- Writing 100 facts into Qwen's n-gram Engram table with zero weight changes: 84% recall — Electronic_Put4530 · 2026-09-21
- Boltzbit and Cambridge team compile context into LoRA weights via hypernetwork, ditching prompt re-reading — xiaohu · 2026-09-21
- A 3-month curated paper list for learning distributed LLM training and inference — East-Muffin-6472 · 2026-09-21
- FrogNano trains a 4B coding agent to 61.5% SWE-bench via online task synthesis — rohanpaul_ai · 2026-09-21