Figure 4

Motif finder assessment. A: We employ three different strategies to extract motif candidates from genome sequences. The statistical significance of a motif's presence has been tested on an expression data set containing 270 conditions. Motif sets have been reported by at least one approach for 214 conditions at a 5% FDR level. The distribution of the significant motif sets from all discovery pipelines is represented by the Venn diagram. B: Pairwise similarity comparison of motif sets from 159 expression conditions that are covered by predictions from all motif discovery pipelines. The scatterplot shows the distribution of 159 condition-specific average similarity values for each pairwise comparison of motif discovery strategies.