Significance ranking of TFBSs in respective clusters. (a) The TFBSs in the sequences of each cluster were sorted and ranked by ANOVA analysis to determine those sites that best discriminated the different clusters. The TFBSs in each cluster were then assigned ranks (1–164) according to their significance (p-value) from the ANOVA analysis. Highest ranking in red, lowest in blue. (b) Partitioning of gene promoter composition with more stringent PWM matrix similarity thresholds reduces the number of clusters identified by K-means analysis. Shown are four of six clusters containing greater than 4 genes. (groups 1, 2, 5 and 6). (c) Analysis of the most discriminating TFBSs in the four clusters in Figure 2b by ANOVA, as in Figure 2a.
McNutt et al. BMC Bioinformatics 2005 6:259 doi:10.1186/1471-2105-6-259