Figure 1.

Scoring exact patterns over-representation in a selected set of gene promoters. For each SELGi, the corresponding SELGPi is divided in overlapping sequence windows. Approximate patterns occurring in at least 30% of the considered promoters are used to select the corresponding group of exact patterns. Then, within the groups of exact patterns, only patterns showing more occurrences than expected by chance: their total number of occurrences is used to compute over-representation P-value of the whole group of exact patterns. If the P-value is significant after Benjamini FDR correction, they are clustered together with other over-represented patterns that identify over-represented motifs corresponding to the specific SELGPi.