Protein Structure-Function Relationship: A Kernel-PCA Approach for Reaction Coordinate Identification.

Area of Science:

• Biophysics
• Computational Biology
• Machine Learning

Background:

• Understanding protein structure-function relationships is crucial for drug discovery and protein engineering.
• High-dimensional data from molecular dynamics (MD) simulations present challenges for traditional analysis methods.

Purpose of the Study:

• To develop a novel Kernel-PCA model for analyzing protein dynamics and structure-function relationships.
• To identify and rank reaction coordinates based on their impact on protein properties.
• To uncover correlations in residue-level dynamics associated with specific protein functions.

Main Methods:

• Kernel-PCA model applied to high-dimensional protein data from MD simulations.
• Ranking of reaction coordinates by their influence on protein properties.
• Residue-level dynamical network analysis to identify correlated dynamics.

Main Results:

• The Kernel-PCA model effectively captures structure-function relationships in proteins.
• Identified and ranked key reaction coordinates impacting protein properties.
• Demonstrated model effectiveness on a G protein-coupled receptor.
• Uncovered residue dynamics correlations linked to specific protein properties.

Conclusions:

• The proposed Kernel-PCA model is a powerful tool for protein structure-function analysis.
• The model facilitates visualization and understanding of complex protein dynamics.
• This approach offers insights into the molecular mechanisms underlying protein function.