Integration with network information underpins the association of MCE with cell potency. (A) Violin plots of the observed MCE values for NPCs and terminally differentiated neurons (scRNA-Seq data from Wang et al.). P-value derives from a one-tailed Wilcoxon rank sum test. (B) As (A), but now for a random permutation of the expression values over the network, leading to a reduced discrimination accuracy. (C). Comparison of the AUC discrimination measure for the observed case [depicted in (A)] and for 100 distinct permutations. P-value is an empirical one comparing the observed AUC to those from the 100 permutations. (D) MCE (y-axis) is approximated reasonably well by the Pearson Correlation Coefficient (PCC) between the transcriptome and connectivity/degree profile of the network (x-axis). |$R^2$| value is given. (E) Boxplots of t-statistics of differential expression between NPCs and neurons (y-axis) against node degree (x-axis), with larger degrees binned into equal-sized groups. Green dashed line denotes the line t = 0. Red line is that of a linear regression. PCC is the Pearson correlation between the t-statistics and node degree. We also give the P-value for the linear regression.