2.4 Protein-Protein Interaction (PPI) network analysis

We applied the online database STRING (https://string-db.org/) to perform the PPI network analysis of evaluating the protein-protein interactions between the screened DEGs. The STRING, a PPI database, collects and assesses evidence from many sources, such as scientific literature, to integrate all known and predicted associations between proteins, including physical interactions and functional associations [13]. All PPI pairs with an interaction score of >0.9 as the threshold value were extracted. Moreover, we used Cytoscape v3.7.2 software plugin cytoHubba to calculate the degree of all nodes and construct PPI networks through different topological analyses. Cytoscape, an open source software project that integrates biomolecular interaction networks with high-throughput expression data and other molecular states into a unified conceptual framework, is one of the most powerful network biology analyses and visualization tools when applied in conjunction with large databases of protein-protein, protein-DNA, and genetic interactions [14, 15]. CytoHubba was composed by Chia-Hao Chin et al. for ranking and exploring important nodes in biological networks by their network features, which provides 11 topological analysis methods in one-stop shopping way [16]. The proteins with higher degrees are more likely to be essential proteins, and the importance of nodes within a biological network will be evaluated. These node ranking methods can be divided into two categories: local and global methods. The local-based methods are considered to be the better methods in discovering essential proteins, which only focus on the relationship between the node and its direct neighbors. Due to the heterogeneity of biological networks, it is reasonable to implement multiple methods for capturing essential proteins [16]. Thus, the genes with the top 10 highest degree values were screened by the 3 topological analyses, including maximal clique centrality (MCC), maximum neighborhood component (MNC), and Degree, all of which are local-based methods. The overlapping hub genes in the top 10 by these 3 topological methods were identified.