MISCORE: a new scoring function for characterizing DNA regulatory motifs in promoter sequences

BMC Systems Biology

Table 1 Conservation and rareness characterization of functional motifs

TF	R(M_t)	Conserved (M_c) models 5000 models			Random (M_r) models 5000 models
		E{R(M_c)} ± std	p-value	z-score	E{R(M_r)} ± std	p-value	z-score
CREB	0.188	0.257 ±_0.025	0.009	02.75	0.458 ±_0.016	0.000	16.60
SRF	0.193	0.286 ±_0.025	0.000	03.76	0.458 ±_0.012	0.000	22.01
TBP	0.134	0.243 ±_0.027	0.000	04.04	0.493 ±_0.008	0.000	43.79
MYOD	0.104	0.195 ±_0.036	0.004	02.54	0.467 ±_0.016	0.000	22.22
ERE	0.214	0.331 ±_0.012	0.000	10.15	0.439 ±_0.007	0.000	31.87
E2F	0.203	0.309 ±_0.019	0.000	05.65	0.444 ±_0.009	0.000	27.54
CRP	0.307	0.380 ±_0.006	0.000	11.48	0.422 ±_0.005	0.000	21.45
GAL4	0.246	0.261 ±_0.016	0.181	00.88	0.418 ±_0.008	0.000	20.95
CREB*	0.188	0.224 ±_0.024	0.058	01.47	0.460 ±_0.017	0.000	15.76
SRF*	0.193	0.261 ±_0.023	0.000	03.01	0.461 ±_0.010	0.000	26.46
TBP*	0.134	0.186 ±_0.026	0.010	02.03	0.491 ±_0.007	0.000	48.37
MYOD*	0.104	0.158 ±_0.033	0.057	01.62	0.472 ±_0.015	0.000	24.05

Remark: the following relation R(M_t) <E{R(M_c)} <E{R(M_r)} indicates the characterization of the conservation property by MISCORE, while the rareness is indicated by a smaller p-value and a larger z-score obtained by the R(M_t) models (true models) compared to the R(M_c) (conserved) and R(M_r) (random) models. Here, z-score(M_t, M_r) = [E{R(M_r)} - R(M_t)]/std{R(M_r)}, and p-value(M_t, M_r) = n/5000, where n is the number of the random models that can hold R(M_r) ≤ R(M_t). It reads similarly for the conserved models M_c. E{*} is the mathematical expectation. Note: Datasets with asterisk are composed of promoters with 500bp, while the others have 200bp in length.

ISSN: 1752-0509