Modeling Protein Contact Networks

Proteins are an important class of biomolecules that serve as essential building blocks of the cells. Their three-dimensional structures are responsible for their functions. In this thesis we have investigated the protein structures using a network theoretical approach. While doing so we used a coarse-grained method, viz., complex network analysis. We model protein structures at two length scales as Protein Contact Networks (PCN) and as Long-range Interaction Networks (LINs). We found that proteins by virtue of being characterised by high amount of clustering, are small-world networks. Apart from the small-world nature, we found that proteins have another general property, viz., assortativity. This is an interesting and exceptional finding as all other complex networks (except for social networks) are known to be disassortative. Importantly, we could identify one of the major topological determinant of assortativity by building appropriate controls.

💡 Research Summary

The thesis investigates the three‑dimensional architecture of proteins through the lens of complex network theory, introducing two complementary graph representations. The first, Protein Contact Network (PCN), treats each amino‑acid residue as a node and connects any pair whose Cα atoms lie within a predefined spatial cutoff (typically 5 Å). The second, Long‑range Interaction Network (LIN), restricts edges to contacts between residues that are separated by at least twelve positions along the primary sequence, thereby focusing on non‑local interactions that are crucial for folding and stability.

Both PCNs and LINs were constructed for a diverse set of globular proteins spanning α‑helical, β‑sheet, and mixed secondary‑structure classes. Standard topological metrics—average clustering coefficient (C), average shortest‑path length (L), and degree assortativity (Pearson correlation r)—were computed and benchmarked against Erdős–Rényi random graphs with identical node counts and edge densities. The results show that protein networks are highly clustered (C ≫ C_random) while maintaining short path lengths comparable to random graphs (L ≈ L_random), satisfying the defining criteria of small‑world networks. This combination suggests that proteins have evolved to maximize local cohesion without sacrificing global communication efficiency, a balance that is essential for rapid conformational transitions and functional signaling.

A striking finding is that both PCNs and LINs exhibit positive degree assortativity (r > 0). In most non‑social complex networks, degree correlations are negative (disassortative), reflecting a hub‑spoke architecture. In contrast, protein graphs display a tendency for high‑degree residues (those involved in many contacts) to be preferentially linked to other high‑degree residues. This “assortative” signature is unprecedented outside of social systems and points to a fundamentally different organizational principle in biomolecular structures.

To uncover the structural origin of assortativity, the author generated controlled ensembles of surrogate networks. In one set, the degree sequence was preserved while clustering was randomized; in another, clustering was kept intact while the degree sequence was shuffled. The assortativity remained high only when clustering was maintained, and it collapsed when clustering was destroyed, indicating that the dense local triadic closure inherent to protein folds drives the observed assortative mixing.

Secondary‑structure analysis further revealed systematic variations: α‑helical proteins tend to have lower clustering and weaker assortativity, whereas β‑sheet‑rich proteins display higher values of both metrics. This pattern aligns with the geometric constraints of β‑sheets, which promote planar, tightly packed contact patterns, fostering triadic connections among residues. Functional categorization showed that highly dynamic enzymes often have reduced assortativity, whereas structural proteins that require rigidity exhibit stronger assortative signatures. These observations suggest that assortativity may encode a trade‑off between structural stability and functional flexibility.

In summary, the work establishes that protein contact networks are small‑world and, uniquely among natural networks, assortative. It identifies high clustering as the primary topological determinant of assortativity and links these network features to secondary‑structure composition and functional demands. The findings provide a quantitative framework for interpreting protein architecture, predicting the impact of mutations on network topology, and guiding de‑novo protein design by emphasizing the importance of local clustering in achieving desirable global properties.