Information theory optimization of signals from small-angle scattering measurements

Area of Science:

• Biophysics
• Structural Biology
• Biophysical Chemistry

Background:

• Small-angle X-ray scattering (SAXS) is crucial for studying biological macromolecules in solution.
• Current bioSAXS methods face resolution limitations, particularly in analyzing particle distribution (P(r)-distribution).
• The inverse transform of SAXS data has been historically considered ill-posed, necessitating indirect analysis.

Purpose of the Study:

• To re-evaluate the inverse transform of SAXS data using matrix and information theories.
• To demonstrate that the inverse problem is well-conditioned and directly related to the Shannon number.
• To develop a quantitative framework for assessing SAXS data quality and model fitting.

Main Methods:

• Application of matrix and information theories to SAXS data analysis.
• Exploitation of oversampling from modern detectors for direct inverse Fourier transforms.
• Development of a hybrid scoring function incorporating Akaike information criteria and Durbin-Watson statistic.

Main Results:

• The inverse transform of SAXS data is shown to be a well-conditioned problem, not ill-posed.
• The Shannon number defines the fundamental resolution limit and maximum information recoverable from SAXS data.
• A novel hybrid scoring function provides a robust assessment of model-data fit and P(r)-distribution quality.

Conclusions:

• Modern detectors and information theory enable direct, well-conditioned inverse transforms of SAXS data.
• The Shannon limit provides a quantitative measure of resolution in bioSAXS experiments.
• The developed framework enhances the precision of determining particle dimensions and selecting appropriate structural models for bioSAXS data.