Table 2.

Assigning TF roles for different CRMs.

CRM	Bcd	Cad	Hb	Tll	Gt	Kr	Kni	Torre
Kr_CD1_ru	- (Δ)	+ (∪)	- (∪)	- (β)	- (β)	+	-	NA
eve_37ext_ru	-	+	+ (β)	+	+ (Δ)	NA	-	-
eve_stripe2	+	+ (Δ)	- (Δ)	-	-	-	-	NA
hb_anterior_actv	+	NA	+ (β)	-	- (Δ)	-	-	NA
kni_+1	+ (β)	- (Δ)	- (β)	-	+	- (Δ)	+	-
run_stripe5	+	+	-	-	-	-	-	- (β)
roles	s	+ (∪)	s (β)	- (∪)	s	s (β)	- (Δ)	s (β)
confidence	111	110	127	94	110	117	103	104

The first rows give the predictions for the TF roles for six of the 44 CRMs (see Additional file 1 Section 1 for the complete set of predictions). The roles for each TF (columns) are determined by majority vote of the three methods: "∪" - SMALLEST-OPTIMAL, "β" - BEST-N and "Δ" - SENSITIVITY. "+" means activator, "-" repressor, respectively and "NA" indicates that no "strong" role prediction could be made for the CRM. The disagreeing method, if any, is shown in brackets. The second last row gives the overall prediction of the role of the TF: activator, repressor or switcher, "s" (see main text for method). The last row shows the number of "strong" role predictions summed over the 44 CRMs and three methods (total of 132).