Combining small angle X-ray scattering (SAXS) with protein structure predictions to characterize conformations in...

Area of Science:

• Structural biology
• Computational biology
• Biophysics

Background:

• Machine learning advances enable accurate protein structure predictions.
• Experimental Small Angle X-ray Scattering (SAXS) data offers insights into protein structures in solution.
• Limitations in computational predictions can be addressed by integrating experimental data.

Purpose of the Study:

• To describe methods for obtaining and comparing protein structure predictions with experimental SAXS data.
• To demonstrate how SAXS data can refine computational models for improved accuracy and biological relevance.
• To explore the potential of experimentally-validated predictions for functional annotation and identifying conserved sites in metagenomic data.

Main Methods:

• Generating protein structure predictions using machine learning algorithms.
• Acquiring and analyzing experimental Small Angle X-ray Scattering (SAXS) data.
• Comparing computational predictions with SAXS data to identify and correct discrepancies.
• Refining atomic models to be consistent with experimental solution data.

Main Results:

• Integration of SAXS data refines protein structure predictions.
• Improved models accurately reflect protein structural information in solution.
• Experimentally-validated predictions enhance functional annotation for metagenomic data.
• Identification of functional clustering on conserved sites is improved, even with low sequence homology.

Conclusions:

• Combining machine learning predictions with SAXS data validation yields more accurate and biologically relevant protein models.
• This integrated approach has broad potential for improving functional annotation and understanding protein families in large-scale genomic datasets.