Predicting protein-nucleic acid flexibility using persistent sheaf Laplacians

Area of Science:

• Biophysics
• Computational Biology
• Structural Biology

Background:

• Protein-nucleic acid complex flexibility, measured by atomic B-factors, is crucial for understanding structure, dynamics, and function.
• Traditional models like Gaussian Network Models (GNM) and Elastic Network Models (ENM) struggle with multiscale interactions in large biomolecular systems.

Purpose of the Study:

• To apply the Persistent Sheaf Laplacian (PSL) framework for accurate B-factor prediction in protein-nucleic acid complexes.
• To evaluate PSL's performance against established methods on diverse biological datasets.

Main Methods:

• Utilized the Persistent Sheaf Laplacian (PSL) framework, integrating multiscale analysis, algebraic topology, combinatorial Laplacians, and sheaf theory.
• Benchmarked PSL against GNM and multiscale FRI (mFRI) on protein-RNA and nucleic-acid-only structures.

Main Results:

• PSL demonstrated superior B-factor prediction accuracy compared to GNM and mFRI.
• Achieved up to a 21% improvement in Pearson correlation coefficient for B-factor prediction.
• PSL effectively captures topological invariants and homotopic shape evolution in biomolecular data.

Conclusions:

• The PSL framework provides a robust and adaptable method for modeling complex biomolecular interactions.
• PSL's accuracy in B-factor prediction highlights its potential for applications in mutation impact analysis and drug design.