Fig. 3

Graph representations of networks of genes with correlated expression profiles. Nodes represent genes, identified by their SACE numbers, and edges are drawn between genes whose expression levels show a strong (positive or negative) correlation in different contexts. a Edges are drawn between genes whose expression levels show a strong correlation in the WT strain and maintain the same correlation in the HP strain. Only genes closely correlated to at least one mutated gene (p < 0.00017 in WT) are shown. The ery cluster is connected to both the primary carbon metabolism (including mutB/SACE_5639, highlighted in magenta) and secondary metabolic genes. A separate cluster of ribosomal proteins was also identified using this approach, but it is not correlated to ery genes. b Edges are drawn between genes whose expression is strongly correlated in the full dataset, across both strains. Only genes with a strong correlation to the ery cluster (top 6 % of all genes) and a large change in expression between WT and HP strains (top 3.7 %) are shown. Genes of the pks6 cluster (SACE_4561-76) form a strongly interconnected group that is overexpressed in HP and includes a mutated gene (SACE_4563) and ilvB (SACE_4565, highlighted in magenta). c Edges are drawn between genes that are strongly correlated in the WT strain but whose correlation is greatly perturbed in the HP strain. Only genes with a high WT correlation to at least one mutated gene (p < 0.00016) and with a large change in expression between HP and WT (top 3.7 %) are shown. The putative sigma24 factor (SACE_5521, highlighted in magenta) loses many positive correlations to pks2 and pks8 genes. d Correlations between the ery cluster and regulatory genes. Edges are drawn between genes that maintain strong correlations in both strains. Genes consistently correlated with ery represent potential activators and those anticorrelated represent potential repressors