Figure 2
From: Estimating the similarity of alternative Affymetrix probe sets using transcriptional networks
Distribution of probe set pair properties (m27-1p). A – Frequency of probe set pairs that belong to the SS class and are positively correlated in 1 to 21 networks. B – Distribution of positive correlations between alternative probe sets. C – Distribution of negative correlations between alternative probe sets. D – Distribution of the neighbourhood overlap p-values of alternative probe sets. In B, C and D, there are 21 curves, one for each subset of alternative probe sets that were positively correlated in 1 to 21 networks (change from cool to warm colours indicates that the number of networks is progressively increasing). The curves corresponding to the alternative probe sets that were positively correlated in all networks and their 5th (B) or 95th (C,D) percentiles (which are used as limits to test whether alternative probe sets must be considered as similar in a given network) are indicated by arrows.