Link clustering explains non-central essential genes in protein interaction networks

Essential genes (EGs) often form central nodes in protein-protein interaction (PPI) networks. However, many reports have shown that numerous EGs are non-central, suggesting that another principle governs gene essentiality. We propose link clustering as a distinct indicator of the essentiality for non-central nodes. Specifically, in various human and yeast PPI networks, we found that 29 to 47% of EGs were better characterized by link clustering than by centrality. Such non-central EGs with clustered links have significant impacts on communities at lower hierarchical levels, suggesting that their essentiality derives from functional dependency among relevant local neighbors, rather than their implication on global connectivity. Moreover, these non-central EGs exhibited several distinct characteristics: they tend to be younger and fast-evolving, and likely change their essentiality across different human cell lines and between human and mouse than central-EGs. Author summaryThe centrality in network structure was thought to be the cause of gene essentiality, as it conveys nodes importance in given networks. However, many essential genes are also found to be non-central, which leads us to ask whether a principle other than centrality may govern gene essentiality. Here we demonstrate that link clustering conveying functional dependency between nodes is the other principle characterizing gene essentiality. The link clustering explains numerous essential genes that are non-central, which are distinct from central ones regarding their molecular function and evolution. Our results clearly establish a two-dimensional principle governing gene essentiality.