Figure 2

Genome wide prediction of Pho regulons in the 19 sequenced cyanobacterial genomes. Green lines represent the probability (p($S_{I_U}$ > s)) that putative three tandem Pho boxes found in the promoter region $I_{U(g_1,\mathrm{...},g_n)}$ of an operon U(g₁,..., g _n ) have a score greater than s (s is a positive number); blue lines represent the probability (p($S_{C_U}$ > s)) that putative three tandem Pho boxes found in a randomly chosen coding region $C_{U(g_1,\mathrm{...},g_n)}$ with the same length as $I_{U(g_1,\mathrm{...},g_n)}$ have a score greater than s (s is a positive number); Red lines represent the log-odds ratio function defined as LOR(s) = ln(p($S_{I_U}$ > s)/p($S_{C_U}$ > s)). The doted vertical line in each panel shows the value of s, such that p($S_{C_U}$ > s) <0.01, and s is the cutoff for SphR binding site predictions. p($S_{C_U}$ > s) is also used to estimate the p value.