Table 4 gives average degrees of essential and nonessential proteins in the raw yeast PPI networks.
Table 4 also shows the average degrees of essential and nonessential proteins in the HC yeast PPI networks together with P-values from two-sample KS tests.
Therefore, in order to determine the significance of the difference between degrees of essential and nonessential proteins we use the two-sample KS test to compare their degree distributions.
It has been previously reported that the Ito data set has a weak correlation between degree and essentiality [20]; however, we find no difference between degree distributions of essential and nonessential proteins for this data set.
Then we visually compared the degree distributions of essential and nonessential membrane proteins to determine if the degrees are drawn from the same underlying population.
Correlations between essentiality and degree were evaluated by calculating P-values for two-sample Kolmogorov-Smirnov (KS) test for differences between degree distributions of essential and nonessential proteins.
We find, through strict statistical analyses of degree distributions of essential and nonessential proteins, that the raw and HC Y2H data sets show no correlation between degree and essentiality, while the HMS-PCI and TAP PPI networks, with one exception, contain substantial correlations.
The relationship between mRNA decline and growth rate decline reflects the degree of essentiality, or stringency, of an essential gene, which is here defined by the minimum transcript level for a 50% reduction in growth rate (MTL50).
This ensures that artificial differences arising from changes in degree distributions of essential proteins are not observed whilst simultaneously conserving the network structure.
In stark contrast, we find that degree distributions of essential and nonessential proteins in the raw Y2H networks are not significantly different, with PKS-values of 0.9915 and 0.1542 for the raw Ito and Uetz data sets, respectively.
