| Summary |
The interpretation of small and wide-angle X-ray scattering (SWAXS) data from biomolecular samples has remained challenging for multiple reasons. First, SWAXS curves contain only few independent data points, whereas biomolecules contain many more degrees freedom. Second, since scattering contributions from the solvent (hydration layer and excluded solvent) contribute to the signals, such contributions are frequently fitted against the experimental data, thus further reducing the available experimental information. Third, no procedure has been designed that provides statistically founded confidence intervals for the models fitted against the experimental data. These factors lead to a significant risk of overfitting and, hence, of drawing unfounded conclusions.
We aim to overcome such problems by combining SWAXS data with explicit-solvent molecular dynamics (MD) simulations and Bayesian inference. Specifically, we have developed methods for SWAXS curve predictions and for structure refinement fully based on explicit-solvent MD [1,2]. The simulations provide an accurate physical model for the hydration layer and excluded solvent, thus avoiding any solvent-related fitting parameters. In addition, MD simulations restrain the biomolecule into conformations of acceptable free energy, thereby strongly reducing the risk of overfitting the data. The methods are illustrated using systems of proteins, protein/detergent complexes, and detergent micelles [1,5]. More recently, simulations were used to interpret femtosecond time-resolved SWAXS data following CO photodissociation in myoglobin [4], and we outline developments for Bayesian structure refinement [4].
[1] Chen and Hub, Biophys. J. 107, 435-447 (2014); Chen and Hub, Biophys. J. 108, 2573–2584 (2015); Chen and Hub, J. Phys. Chem. Lett., 6, 5116–5121 (2015)
[2] http://waxsis.uni-goettingen.de; Knight and Hub, Nucleic Acids Res., 43, W225-W230 (2015)
[3] Brinkmann and Hub, PNAS 113, 10565–10570 (2016); Brinkmann and Hub, J. Chem. Phys., 143, 104108 (2015)
[4] Shevchuk and Hu
|
