Reliable Biomolecular Structural Modelling with Small-Angle Scattering

Abstract

The small-angle scattering (SAS) of X-rays or neutrons from biomolecules in solution can provide (1) unique structural insights, (2) information on the range of potential conformations present, (3) and powerful restraints for integrative/hybrid structural modelling (IHM). The rotational and ensemble averaging of the signal presents multiple challenges to the reliable utilisation of SAS data: (1) demonstrating SAS profile represents the species of interest, (2) reliably evaluating model fits, and (3) over-interpretation and over-fitting of data. A community driven project has been underway for more than a decade now to define the information that should be made available regarding sample quality, data acquisition and reduction, presentation of scattering data and validation, and modelling for biomolecular SAS experiments. These guidelines (Trewhella et al. (2017) Acta D73, 710) provide a reporting framework so that “readers can independently assess the quality of the data and the basis for any interpretations presented” concerning the results of bimolecular SAS experiments, including the generation or testing of 3D models. The validation task forces for SAS (Trewhella et al. (2013) Structure 21, 875) and for IHM (Sali et al. (2015) Structure 23, 1156) of the world-wide Protein Data Bank are additionally addressing the complex issues concerning the archiving and validation of IHM models that depend upon computational methods and disparate data from multiple techniques (Burley et al. (2017) Structure 25, 1317). Finally, the question of how many parameters are justified by the model evidence inherent in the SAS profile is an important question to answer when modelling conformational ensembles (Potrzebowski, Trewhella and Andr⌈ (2018) in review PLOS Computational Biology). This presentation will address the recommended reporting framework and its rationale, the reliable incorporation of SAS data in IHM, and the strengths and limits of SAS in ensemble modelling.