The abstract is read at the publisher; the summary is JClub's.
ProtFlap predicts relative residue-level protein flexibility from single static, energy-minimised structures using a graph neural network, achieving robust and accurate performance across diverse protein domains.
- Why it matters: Protein flexibility underpins molecular recognition, catalysis, and regulation, but MD-based characterization is too computationally expensive for scale, and static structure predictors miss dynamics.
- What they did: ProtFlap builds residue interaction graphs and combines structural/sequence descriptors with empirical atomic energy features from one snapshot, training on mobility labels from short MD–derived normalised RMSF.
- The result: The model’s flexibility estimates are robust and accurate across diverse domains, and attribution analysis indicates van der Waals-related energy terms dominate the predictions.
The findingWhy it mattersWhat they didThe result
AI summary of the abstract · may contain errors · not medical advice · read the paper.