Fig. 2

Gene expression network modeling depending on ERα36 expression level. a Network distance characterization as a function of ERα36 expression level (see text for details). Expression level varied between 0 and 20 in the samples expressing ERα36 (x-axis). With step of 0.5 on ERα36 expression level, population was divided into two sub-groups for which networks were computed. A distance between corresponding networks was calculated (y-axis). The ERα36 expression threshold corresponding to the most different gene networks was computed and was equal to 8.35. b–c Graphs were designed by computing nonlinear correlation and mutual information between each gene expression pair in either low ERα36 [ER α36+] expressing (b) or high ERα36 [ERα36++] expressing (c) samples. The vertices represent genes. The edges linking the vertices indicate that independence between gene expressions is less than 0.05. Positive correlations are in blue and negative correlations in red. Correlation values are given in Additional file 3: Table S2A and Additional file 4: Table S2B and P-values in Additional file 5: Table S3A and Additional file 6: Table S3B, respectively