Characterization of degree frequency distribution in protein interaction networks

In this work, we analyze the degree frequency distribution in the yeast protein interaction network by studying a previously proposed duplication network model. This model correctly predicts the observed degree distribution (a power law for large degree values and a departure from this behavior for small degree). We numerically and analytically characterize this distribution as a mixture of random and power-law behavior, and make a comparative study of the robustness of the network model against realistic perturbations. We conclude that the particular distribution observed in both the model and the experimental data has many advantages in terms of dynamical and topological robustness and could have emerged in the evolutionary history as a sort of compromise between purely deterministic and random underlying mechanisms of network growth.