Revisiting topological properties and models of protein-protein interaction networks from the perspective of dataset evolution

Abstract

Protein-protein interaction (PPI) networks are crucial for organisms. Many research efforts have thus been devoted to the study on the topological properties and models of PPI networks. However, existing studies did not always report consistent results on the topological properties of PPI networks. Although a number of PPI network models have been introduced, yet in the literature there is no convincing conclusion on which model is best for describing PPI networks. This situation is primarily caused by the incompleteness of current PPI datasets. To solve this problem, in this study, the authors propose to revisit the topological properties and models of PPI networks from the perspective of PPI dataset evolution. Concretely, they used 12 PPI datasets of Arabidopsis thaliana and 10 PPI datasets of Saccharomyces cerevisiae from different Biological General Repository for Interaction Datasets (BioGRID) database versions, and compared the topological properties of these datasets and the fitting capabilities of five typical PPI network models over these datasets.

Fig 1

Number of nodes/edges changes with the version of A. thaliana dataset a Number of node/edges against dataset version b Number of edge against number of node

Fig 2

Degree distributions of different A. thaliana datasets a Version 65 b Version 85 c Version 110

Fig 3

Topological properties of the GCs on the A. thaliana datasets of different versions a Average degree b Betweeness c 2/3/4‐hop reachability d Closeness e Average shortest path f Diameter g Clustering coefficient h Eccentricity

Fig 4

Ranks of the five models (BA, ER, ERDD, GEO and STICKY models) on different PPI data versions of a A. thaliana b S. cerevisiae Here, the X ‐axis denotes dataset version and the Y ‐axis denotes model rank by MRR