Papers

Quantitative and Predictive Folding Models from Limited Single-Molecule Data Using Simulation-Based Inference

L Dingeldein, A Lyons, P Cossio, M Woodside… - arXiv preprint arXiv …, 2025 - arxiv.org
Physics paper physics.chem-ph Suggest

… Here, we introduce a framework based on simulation-based inference (SBI) that overcomes these limitations by integrating physics-based modeling with deep learning. …

Link to paper

BibTeX

@article{2508.02509v1,
Author        = {Lars Dingeldein and Aaron Lyons and Pilar Cossio and Michael Woodside and Roberto Covino},
Title         = {Quantitative and Predictive Folding Models from Limited Single-Molecule
  Data Using Simulation-Based Inference},
Eprint        = {2508.02509v1},
ArchivePrefix = {arXiv},
PrimaryClass  = {physics.chem-ph},
Abstract      = {The study of biomolecular folding has been greatly advanced by
single-molecule force spectroscopy (SMFS), which enables the observation of the
dynamics of individual molecules. However, extracting quantitative models of
fundamental properties such as folding landscapes from SNFS data is very
challenging due to instrumental noise, linker artifacts, and the inherent
stochasticity of the process, often requiring extensive datasets and complex
calibration experiments. Here, we introduce a framework based on
simulation-based inference (SBI) that overcomes these limitations by
integrating physics-based modeling with deep learning. We apply this framework
to analyze constant-force measurements of a DNA hairpin. From a single, short
experimental trajectory of only two seconds, we successfully reconstruct the
hairpin's free energy landscape and folding dynamics, obtaining results that
are in close agreement with established deconvolution methods that require
approximately 100 times more data. Furthermore, the Bayesian nature of this
approach robustly quantifies uncertainties for inferred parameter values,
including the free-energy profile, diffusion coefficients, and linker
stiffness, without needing independent measurements of instrumental properties.
The inferred model is predictive, generating simulated trajectories that
quantitatively reproduce the thermodynamic and kinetic properties of the
experimental data. This work establishes SBI as a highly efficient and powerful
tool for analyzing single-molecule experiments. The ability to derive
statistically robust models from minimal datasets is crucial for investigating
complex biomolecular systems where extensive data collection is impractical or
impossible. Consequently, our SBI framework enables the rigorous quantitative
analysis of previously intractable biomolecular systems, paving the way for
novel applications of SMFS.},
Year          = {2025},
Month         = {Aug},
Url           = {http://arxiv.org/abs/2508.02509v1},
File          = {2508.02509v1.pdf}
}

Share