Graph theoretical studies on protein structure, folding and stability

Abstract

Graph Spectral Analysis of Protein Structures: Insights into Folding, Stability, and Function

Abstract and Synopsis Introduction Efficient analysis of protein 3D structures provides insights into their function, folding, and stability. Clusters of side-chain interactions in native protein structures are often formed early during folding, with charged clusters near active/binding sites and hydrophobic clusters contributing to protein-protein, protein-DNA interactions, and internal stability. While earlier studies identified such clusters, a systematic and efficient method was needed. This thesis develops a novel algorithm based on graph theory to analyze protein structures.

Methodology (Chapter 2)

Protein structures are represented as graphs (nodes = residues, edges = interactions).

Graph spectral theory (eigenvalues and eigenvectors of graphs) is applied to identify residue clusters.

Unlike local methods, this approach captures global non-bonded interactions.

A single numeric computation identifies clusters and cluster centers (residues with maximum interactions).

Applications

Chapter 3: Identifying conserved clusters among topologically similar proteins, transition-state interactions during folding, and domain identification.

Chapter 4: Analysis of thermophilic proteins revealed additional aromatic clusters compared to mesophilic homologues, providing new insights into thermal stability.

Chapter 5: Application to RNA polymerase assembly identified a nine-residue cluster in the N-terminal domain of the -subunit. Predicted mutations disrupting dimerization were experimentally validated.

Chapter 6: Cluster analysis of 36 / barrel enzymes identified conserved stabilizing interactions in -barrel regions. Cluster centers occurred near active sites, predicting folding nuclei.

Chapter 7: Development of new parameters to discriminate native from non-native structures using spectral properties.

Chapter 8: Spectral parameters efficiently distinguished native folds from decoy structures, aiding protein model evaluation.

Chapter 9: Backbone packing density analysis revealed amino acid preferences across structural classes, presented in novel graphical formats.

Conclusions

Graph spectral methods provide a powerful tool for analyzing protein folding, stability, and function.

Applications include predicting active/binding sites, analyzing protein-protein and protein-DNA interfaces, and evaluating predicted protein models.

The presence of aromatic clusters in thermophilic proteins and conserved clusters in / barrel enzymes highlight structural determinants of stability.

This approach offers new directions for protein engineering, folding simulations, and comparative modeling.