Network motif identification in stochastic networks

Jiang et al. 10.1073/pnas.0507841103.

Supporting Information

Files in this Data Supplement:

Supporting Figure 4
Supporting Figure 5
Supporting Figure 6
Supporting Figure 7
Supporting Figure 8
Supporting Table 4
Supporting Table 5
Supporting Table 6
Supporting Text

Supporting Figure 4

Fig. 4. The scheme of swapping two edges.

Supporting Figure 5

Fig. 5. Scheme of changing an element from 0 to 1.

Supporting Figure 6

Fig. 6. Simulation results for recovering 3-node motifs in pseudo regulatory networks. (A) The relationship of the log pseudo-likelihood ratio versus l . (B) The relationship of the p value versus l . (C) The relationship of the relative error el versus l . (D) The relationship of the relative error eQ ₁ versus l .

Supporting Figure 7

Fig. 7. The relationship between p values and corresponding interaction probabilities.

Supporting Figure 8

Fig. 8. The iteration procedure of the expectation-maximization algorithm when identifying the 3-node motifs in the S. cerevisiae regulatory network. The x axis is the iteration step, and the y axis is the parameters (l and q ₁) estimated in each iteration step.

Table 4. Transcriptional regulatory networks for E. coli and S. cerevisiae (based on highly reliable data from manually maintained databases)

Species	Nodes	Edges	Ave. (In/Out) Degree
E. coli	423	519	1.23
S. cerevisiae	688	1,078	1.57

Table 5. Transcriptional regulatory network for S. cerevisiae (based on the ChIP-chip data; p value threshold of 0.001)

Species	Nodes	Edges	Avg. (in/out) degree
S. cerevisiae	2,416	4,344	1.80

Table 6. Protein–protein interaction networks for seven species

Species	Nodes	Edges	Avg. degree
E. coli	553	483	1.75
S. cerevisiae	2,614	6,319	4.83
C. elegans	2,638	3,970	3.01
H. pylori	710	1,359	3.83
M. musculus	329	274	1.67
D. melanogaster	7,068	20,815	5.89
H. sapiens	1,065	1,318	2.48