A dynamic network of transcription in LPS-treated human subjects

Table 2 Major clusters formed from the adjusted strength matrix.

Cluster	Num. of genes	Dominant transcription factors	Avg. of pair-wise correlations	P-value
A	18	NFKB1, RELA	0.5742	<0.001*
B	11	STAT1	0.7815	<0.001*
C	5	STAT3	0.5138	0.025
D	17	JUN, FOS	0.4054	0.001
E	6	MYC	0.4350	0.031
F	5	STAT6	0.7546	0.004
G	21	MYC	0.6442	<0.001*

For N genes in a cluster, N(N-1)/2 pair-wise Pearson's correlations on expression were measured and their average was calculated. Next, N genes were randomly selected from a set of ~18,000 human genes and the average pair-wise correlation was calculated from the random gene set. The distribution of the average pair-wise correlation of random genes was re-estimated 1,000 times to generate a null distribution of the average pair-wise correlations. A p-value for each cluster was estimated by counting the number of random gene sets for which the average correlation is larger than the cluster correlation. *No random gene set that exceeded the cluster was found in 1,000 repeats.