Disassortativity in Biological and Supply Chain Networks

Network science has allowed researchers to model complex real world systems as networks in order to identify non trivial topological patterns. Degree correlations (or assortativity) is one such non trivial topological property, which indicates the extent to which nodes with similar degrees tend to pair up with each other. Biological networks have long been known to display anti-degree correlations (disassortativity), where highly connected nodes tend to avoid linking with each other. However, the mechanism underlying this structural organisation remain not well understood. Recent work has suggested that in some instances, disassortativity can be observed merely as a model artefact due to simple network representations not allowing multiple link formations between the node pairs. This phenomena is known as structural disassortativity. In this paper, we analyse datasets from two distinct classes of networks, namely; man made supply chain networks and naturally occurring biological networks. We examine whether the observed disassortativity in these networks are structurally induced or owing to some external process. Degree preserving randomisation is used to generate an ensemble of null models for each network. Comparison of the degree correlation profiles of each network, against that of their degree preserving randomised counterparts reveal whether the observed disassortativity in each network is of structural nature or not. We find that in all biological networks, the observed disassortativity is of structural nature, meaning their disassortative nature can be fully explained by their respective degree distributions, without attribution to any underlying mechanism which drives the system towards disassortativity. However, in supply chain networks, we find one case where disassortativity is structurally induced and in other cases where it is mechanistically driven. We conclude by emphasizing on ruling out structural disassortativity in future research, prior to investigating mechanisms underlying disassortativity in networks