Genome-wide prediction, display and refinement of binding sites with information theory-based models

BMC Bioinformatics

Table 1 Total binding site counts based on genome scans of promoters with PXR/RXRα information weight matrices

Models Compared		Numbers of sites in each category
		Unique sites		Z scores			R _i			Confidence intervals ⁺
A	B	A-B *	B-A^	Threshold (ΔZ)	(A ∩ B) S ^~	(A ∩ B) I ^@	Threshold (Δ R _i , bits)	(A ∩ B) S	(A ∩ B) I	Threshold (± S.D.)	(A ∩ B) S	(A ∩ B) I
1	2	11758	45219	0.5	27945	44302	1	29378	42869	1	30982	41265
				0.75	7492	64755	2	9080	63167	2	26931	45316
				1.0	589	71658	3	2293	69954	3	23625	48622
2	3	17065	157922	0.5	90459	9942	1	54426	45975	1	55431	44970
				0.75	73309	27092	2	26038	74363	2	45069	55332
				1.0	48657	51744	3	11044	89357	3	37822	62579
3	4	61906	148894	0.5	54586	141831	1	93585	102832	1	104397	92020
				0.75	17891	178526	2	33843	162574	2	80088	116329
				1.0	5044	191373	3	11069	185348	3	68846	127571

⁺ Standard error computation for individual R_ivalues is based on derivation given in reference 11; *Sites found with model A but not with model B; ^sites found with model B, but not with model A; ~ Number of sites with differences in R_ivalues exceeding threshold Z scores; ^@Number of sites with differences in R_ivalues less than threshold.

ISSN: 1471-2105