A comprehensive statistical study of metabolic and protein–protein interaction network properties

Understanding the mathematical properties of graphs underlying biological systems could give hints on the evolutionary mechanisms behind these structures. In this article we perform a complete statistical analysis over thousands of graphs representing metabolic and protein–protein interaction (PPI) networks. First, we investigate the quality of fits obtained for the nodes degree distributions to power-law functions. This analysis suggests that a power-law distribution poorly describes the data except for the far right tail in the case of PPI networks. Next we obtain descriptive statistics for the main graph parameters and try to identify the properties that deviate from the expected values had the networks been built by randomly linking nodes with the same degree distribution. This survey identifies the properties of biological networks which are not solely the result of their degree distribution, but emerge from yet unidentified mechanisms other than those that drive these distributions. The findings suggest that, while PPI networks have properties that differ from their expected values in their randomized versions with great statistical significance, the differences for metabolic networks have a smaller statistical significance, though it is possible to identify some drift. •Analysis done over thousands of graphs representing metabolic and PPI networks.•The power-law (scale-free) function poorly describes the graphs degree distributions.•The graph parameters are evaluated and compared to their expected values.•Significant fluctuations are observed between the data and the expected values.•Evolutionary models should drive other characteristics than the degree distributions.