Predicting directional flexibility in proteins
Abstract
Predicting protein dynamics is a long-standing problem in computational structural biology. Often, protein function critically depends on local directed motions, such as hinge movements, catalytic loop rearrangements and domain reorientations, which can be characterized by directional flexibility and correlated structural motions of the protein backbone. While Molecular Dynamics (MD) simulations provide an established but often prohibitively expensive approach, recent deep generative models aim to reduce this cost by directly predicting conformational ensembles, emulating MD. However, due to their large size and the need to generate several states until the derived dynamical properties converge, these models remain expensive. In this work, we propose BackFlip-2: a fast SE(3)-equivariant graph neural network trained to directly predict dynamical descriptors, such as directional backbone flexibility and pairwise dynamic correlations, from an equilibrium structure. In a series of experiments, we show that our model matches the accuracy of substantially larger ensemble generation models while being orders of magnitude faster, and demonstrate that the proposed equivariant architecture is especially well-suited for capturing anisotropic motions in proteins. BackFlip-2 model weights, training and inference code are available at https://github.com/graeter-group/backflip.
Related
- From Insulin Resistance to Mitochondrial Failure: A Staged Model of Impaired Metabolic Flexibility in Schizophrenia
- Grey mouse lemurs, Microcebus murinus, are a relevant model to study gut microbiome flexibility in response to diet changes
- Nutrient stress-driven ACSS1 controls metabolic flexibility and bone formation
- Antibody CDR-H3 loop flexibility: Insights from X-ray crystallography, structural bioinformatics, and the limits of current deep learning methods
- Molecular Flexibility and Electrostatic Interactions Drive the Nanoscale Structure of Coiled-Coil GM130 Protein Condensates
BibTeX
@misc{viliuga2026predicting,
title = {{Predicting directional flexibility in proteins}},
author = {Vsevolod Viliuga and Leif Seute and Matteo Tadiello and Nicolas Wolf and Frauke Gräter and Arne Elofsson},
howpublished = {arXiv},
year = {2026},
month = sep,
doi = {10.48550/arxiv.2609.08474},
eprint = {2609.08474},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2609.08474},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Predicting directional flexibility in proteins
AU - Vsevolod Viliuga
AU - Leif Seute
AU - Matteo Tadiello
AU - Nicolas Wolf
AU - Frauke Gräter
AU - Arne Elofsson
PY - 2026
DA - 2026/09/08
PB - arXiv
DO - 10.48550/arxiv.2609.08474
UR - https://arxiv.org/abs/2609.08474
AB - Predicting protein dynamics is a long-standing problem in computational structural biology. Often, protein function critically depends on local directed motions, such as hinge movements, catalytic loop rearrangements and domain reorientations, which can be characterized by directional flexibility and correlated structural motions of the protein backbone. While Molecular Dynamics (MD) simulations provide an established but often prohibitively expensive approach, recent deep generative models aim to reduce this cost by directly predicting conformational ensembles, emulating MD. However, due to their large size and the need to generate several states until the derived dynamical properties converge, these models remain expensive. In this work, we propose BackFlip-2: a fast SE(3)-equivariant graph neural network trained to directly predict dynamical descriptors, such as directional backbone flexibility and pairwise dynamic correlations, from an equilibrium structure. In a series of experiments, we show that our model matches the accuracy of substantially larger ensemble generation models while being orders of magnitude faster, and demonstrate that the proposed equivariant architecture is especially well-suited for capturing anisotropic motions in proteins. BackFlip-2 model weights, training and inference code are available at https://github.com/graeter-group/backflip.
ER -