Discovering the important proteins through persistent homology in aging protein-protein interaction networks

Abstract

A Protein-Protein Interaction Network (PPIN) is a mathematical model in which every protein is described as a node, and the physical interaction or similar protein expression is considered an edge. Previous studies have shown that PPIN performs various analyses and protein predictions in many aspects, such as essential protein prediction and drug targeting. Numerous centrality measures can provide protein characterization at the node level. However, we still have insufficient network-level identification. In this study, Persistent Homology (PH) is incorporated as an additional network-level measurement to analyze 42 aging PPINs, comprising 22 males and 20 females, aged between 20 and 99. The Vietoris-Rips (VR) filtration was used to capture simplicial complexes before obtaining the persistent barcodes, which are considered the topological representation of a network. The derivation of persistent barcodes, named the Betti Sequence, is calculated for each network, which represents the complexity of the network. Node deletion is performed to assess the change in complexity of the network. The findings reveal a significant change in the Betti sequence after node deletion, indicating that the node is crucial within the network and could potentially serve as a drug target.