Accurate Prediction of Protein Structural Flexibility by Deep Learning Integrating Intricate Atomic Structures and Cryo-EM Density Information
Problem
Computationally predicting protein structural flexibility/dynamics was challenging and existing methods were inaccurate.
Why it matters
Protein dynamics are crucial for understanding biological mechanisms.
What the AI did
RMSF-net, a neural network integrating experimental protein structures and cryo-EM density data, predicts protein dynamics in seconds.
Results
5-fold cross-validation: test correlation 0.746 ± 0.127 (voxel level) and 0.765 ± 0.109 (residue level); inference in a few seconds
Impact
Enables fast protein flexibility prediction that outperforms previous methods on a large-scale dataset.
Record
| Venue | Nature Communications |
|---|---|
| Published | 2024 |
| Work type | article |
| Authors | Xintao Song, Lei Bao, Chenjie Feng, Qiang Huang, Fa Zhang, Xin Gao, Renmin Han |
| Citations | 87 (OpenAlex, at import time) |
| References | 60 |
| Model or method | RMSF-net |
| Organization | not stated in source |
| Primary source | https://doi.org/10.1038/s41467-024-49858-x |
| Open access | gold |
| Keywords | protein dynamics, deep learning, cryo-EM, RMSF-net, structural biology |
Importance rating
significant — thousands to millions affected
Abstract says protein dynamics are crucial for understanding biological mechanisms; a foundational protein-science capability, but no scale of impact is stated.