Automatic Bayesian Weighting for SAXS Data.

Area of Science:

• Structural biology
• Biophysics
• Computational biology

Background:

• Small-angle X-ray scattering (SAXS) is a solution-based technique for studying protein structures.
• Determining protein structures from SAXS data is challenging due to computational costs and protein flexibility.
• Existing methods often involve generating many structures and then filtering them based on SAXS data.

Purpose of the Study:

• To develop an efficient Bayesian model for direct structure determination from SAXS data.
• To characterize the dynamic structural ensemble of proteins using SAXS and molecular dynamics.
• To investigate the structure and function of the PTPN4 protein using this integrated approach.

Main Methods:

• Bayesian modeling of SAXS data to drive Monte Carlo simulations.
• Automatic weighting of SAXS data for efficient structure identification.
• Extensive molecular dynamics simulations to assess model stability and dynamic ensembles.
• Application to a tandem domain protein (PTPN4) with an unstructured linker.

Main Results:

• The Bayesian approach efficiently identifies optimal protein structures by directly using SAXS data.
• Molecular dynamics simulations confirm the stability of the primary conformation and reveal a minor, alternative conformation.
• The study characterizes a dynamic structural ensemble consistent with SAXS data.
• SAXS data provided insights supporting and extending other experimental findings on PTPN4.

Conclusions:

• The developed Bayesian SAXS modeling is an efficient method for structure determination.
• The PTPN4 protein exists in a stable conformation with a minor alternative state.
• The unstructured linker may play a role in maintaining PTPN4 in an inhibited enzymatic state.
• This integrated computational approach enhances the understanding of protein dynamics and function.