SAXSTER

SAXSTER integrates small-angle X-ray scattering (SAXS) data with threading and protein-fold-recognition algorithms to improve template-based protein structure prediction and enable residue-level structural assignments for atomic-level modeling.

Key Features:

• Data integration: Couples raw SAXS data with threading and protein-fold-recognition algorithms to inform template-based predictions.
• Scoring functions: Implements nine distinct matching scoring functions that compare experimental SAXS profiles with template profiles.
• Best-performing score: Uses the logarithm of the integrated correlation score, which demonstrated superior performance in recognizing accurate templates.
• Correlation with TM-score: The logarithm of the integrated correlation score shows a strong correlation with Template Modeling (TM) scores of target structures.
• Improved template recognition: Large-scale protein-fold-recognition experiments revealed significant enhancements in identifying optimal template structures.
• Asymmetric profile performance: For proteins with asymmetric SAXS profile distributions, the average TM-score for top-ranking templates increased by 18% after excluding homologous templates.
• Statistical validation: The reported improvement was statistically significant (p < 10^-9, Student's t-test).
• Applicability to complex shapes: Effective for proteins with irregular global shapes and multi-domain complexes.

Scientific Applications:

• High-resolution structure determination: Enhances use of SAXS data to support atomic-level protein structure modeling.
• Template selection in fold recognition: Improves identification of accurate templates for template-based and threading methods.
• Modeling irregular and multi-domain proteins: Applies to proteins with irregular global shapes and multi-domain complexes where traditional methods may underperform.
• Residue-level assignment from low-resolution data: Facilitates residue-level structural assignments using low-resolution SAXS profiles.

Methodology:

SAXSTER couples raw SAXS data with protein-fold-recognition (threading) algorithms, applies nine matching scoring functions comparing experimental SAXS profiles to template profiles, evaluates performance using the logarithm of the integrated correlation score, and assessed results via large-scale protein-fold-recognition experiments with exclusion of homologous templates and statistical testing by Student's t-test.