2.2 Biological significance of the putative module representatives
Our signal transduction model of Equation 2 provides a vehicle to quantitatively measure the degree of biological and topological influence of each protein on other proteins in the PPI network. The most influential proteins, that is, the highest scored nodes, are highly important proteins. To evaluate the biological significance of the most influential proteins, we annotated the lethality of each protein in the yeast PPI network according to the MIPS lethality data. Lethality is a crucial factor to characterize the biological essentiality of a protein. It is determined by examining whether a module is functionally disrupted when the protein is knocked out. We obtained the protein lethality information from MIPS database [17], which reports whether a protein is lethal or viable. We found that 233 proteins out of the top scored 555 proteins are lethal.
Figure 3 plots the cumulative number of lethal genes vs. the number of protein nodes included for increasing percentiles of the degree, betweeness or the STM signal transduction metric. The data are shown for 555 genes, obtained from the yeast PPI network, with the highest values of each of these metrics. In each case, the results are sorted and highest values are placed closest to the origin. Figure 3 shows that the performance of the STM metric in predicting lethality is comparable to that of degree and betweeness approaches for up to 150 nodes.
Figure 3  Accumulation of lethal proteins for various percentiles. Accumulation of lethal proteins for various percentiles of degree (gray line), betweeness centrality (dashed line) or the STM signal transduction metric (solid line). The results are shown for the top 555 proteins obtained from the yeast PPI network and are ordered; the highest values of these metrics are closest to the origin.