Figure 1

PPIs and protein domain compositions are parsimoniously organized in nature. Under each null model, 200 data sets are simulated. The distribution of T is shown as 'violin plot' and p-value is computed based on Gaussian distribution.