Fine-scale dissection of functional protein network organization by statistical network analysis

Abstract

Revealing organizational principles of biological networks is an important goal of systems biology. In this study, we sought to analyze the dynamic organizational principles within the protein interaction network by studying the characteristics of individual neighborhoods of proteins within the network based on their gene expression as well as protein-protein interaction patterns. By clustering proteins into distinct groups based on their neighborhood gene expression characteristics, we identify several significant trends in the dynamic organization of the protein interaction network. We show that proteins with distinct neighborhood gene expression characteristics are positioned in specific localities in the protein interaction network thereby playing specific roles in the dynamic network connectivity. Remarkably, our analysis reveals a neighborhood characteristic that corresponds to the most centrally located group of proteins within the network. Further, we show that the connectivity pattern displayed by this group is consistent with the notion of "rich club connectivity" in complex networks. Importantly, our findings are largely reproducible in networks constructed using independent and different datasets.

Figure 1. Dissection of proteins into dynamical classes.

Hierarchical clustering of proteins, using Ward's method, by their dynamic profiles (upper panel) and the interaction matrix showing the protein-protein interaction patterns between different dynamic profiles (lower panel). In order to make clustering possible, we normalized each row to have a mean of 0 and a variance of 1. For the lower panel, proteins were binned into 114 bins with the exact ordering as in the clustering in the upper panel. Each square in the matrix represents the number of interactions between respective bins.

Figure 2. Characterization of roles of subgroups in the network connectivity.

a) Deletion profiles of select subgroups. White stripes in the heatmaps indicate the deleted group. Please see Methods for a detailed description of the deletion profiles. b) Normalized rich club coefficients (see Methods) of each group.

Figure 3. The reduced network plot.

Each node in this network represents one of bins used to construct the interaction matrix in Figure 1. Therefore, each bin represents 10 proteins with most similar dynamic profiles. In this network, there is an interaction between two bins only if there are at least 4 number of protein-protein interactions between proteins in the two bins. Bins are colored according to which dynamic class they belong.

Figure 4. Heatmaps for the dynamic profiles of proteins in two independent datasets and their protein-protein interaction profiles.

a) High confidence dataset from Bertins et al (2006). b) High confidence dataset from Batada et al (2006). Ordering of proteins in bins of the interaction matrices are exactly like in the heatmaps above each matrix. Clustering was done the same way as for our dataset (see text).