Most proteins cannot function as single unit but associate subunits via the formation of protein interfaces, to be biologically active. How the amino acids involved in subunit association, so-called hot spots, regulate the formation of a protein interface is still an open question. Here, we show how network and graph theories can help addressing the role of hot spots. We built a MatLab code called SpectralPro which identifies hot spots and reconstructs the protein interface as a subnetwork of hot spots in interaction, with the hot spots as nodes and the bonds between hot spots as links. Using as a case study, the cholera toxin B pentamer (five subunits), we investigate if the degree of a node, namely the number of contacts of a hot spot, is important in the formation of an interface. The degree of a node is known to be important in many real networks. For example in social networks, hubs control the communication between most nodes and as such are vulnerable to changes. But our result shows that in the toxin interface sub-graph hub-like nodes are less vulnerable to change than single link node.