Symmetry in cancer networks identified: Proposal for multi-cancer biomarkers

TL;DR

This study uses spectral graph theory to identify symmetrical patterns in cancer protein networks, proposing new multi-cancer biomarkers based on structural symmetry and functional analysis.

Contribution

It introduces a novel spectral graph approach to detect network symmetry in cancer proteomics data, identifying potential multi-cancer biomarkers.

Findings

Identified structural symmetry in cancer protein networks.

Proposed five proteins as potential multi-cancer biomarkers.

Validated biomarkers through survival analysis.

Abstract

One of the most challenging problems in biomedicine and genomics is the identification of disease biomarkers. In this study, proteomics data from seven major cancers were used to construct two weighted protein-protein interaction (PPI) networks i.e., one for the normal and another for the cancer conditions. We developed rigorous, yet mathematically simple, methodology based on the degeneracy at -1 eigenvalues to identify structural symmetry or motif structures in network. Utilising eigenvectors corresponding to degenerate eigenvalues in the weighted adjacency matrix, we identified structural symmetry in underlying weighted PPI networks constructed using seven cancer data. Functional assessment of proteins forming these structural symmetry exhibited the property of cancer hallmarks. Survival analysis refined further this protein list proposing BMI, MAPK11, DDIT4, CDKN2A, and FYN as putative multi-cancer biomarkers. The combined framework of networks and spectral graph theory developed here can be applied to identify symmetrical patterns in other disease networks to predict proteins as potential disease biomarkers.