Singular value decomposition for the correlation of atomic fluctuations with arbitrary angle.

Area of Science:

• Biophysics
• Structural Biology
• Computational Biology

Background:

• Allostery is a key protein property enabling information transfer between distant sites.
• Allosteric mechanisms are linked to correlated atomic fluctuations within proteins.
• Conventional correlation analysis using dot products underestimates certain fluctuation correlations.

Purpose of the Study:

• To introduce a novel singular value decomposition (SVD) method for analyzing protein fluctuation dynamics.
• To assess the SVD method's ability to capture correlations missed by conventional approaches.
• To evaluate the SVD method's effectiveness in predicting allosteric sites.

Main Methods:

• Developed a singular value decomposition (SVD) method to analyze correlation coefficients of fluctuation dynamics.
• Applied the SVD method to the second PDZ domain (PDZ2) of human PTP1E protein.
• Validated the SVD method on a dataset of 23 known allosteric monomer proteins.

Main Results:

• The SVD method identified significant correlations with near-perpendicular directions, underestimated by conventional methods.
• This underestimation was more pronounced for residue pairs with greater separation.
• SVD analysis showed better agreement with experimentally determined PDZ2 dynamics compared to conventional methods.
• The SVD approach enhanced the prediction accuracy of allosteric sites.

Conclusions:

• The SVD method provides a more comprehensive analysis of protein dynamics and allostery.
• This technique offers improved insights into protein motion and allosteric site identification.
• The SVD method holds potential for advancing protein dynamics research and drug design.