Table 1.

A summary of NRSF ChIP-chip and ChIP-seq binding regions

Data and analysis method	No. of peaks	Peak with NRSF motif	# Motif / 1kb	Region length percentiles (bp)
				10	25	50	75	90
Affy-TileMap	7114	1001 (14.1%)	0.15	211	323	616	1274	2311
Seq-S1w100	3312	1277 (38.6%)	1.26	122	173	269	444	598
Seq-S1w100 (B)	3312	1223 (36.9%)	5.54	29	30	60	82	113
Seq-S1w100 (B+S)	1861	1051 (56.5%)	6.98	41	59	73	90	122
Seq-S2w100	3317	1280 (38.6%)	1.28	116	161	261	445	604
Seq-S2w100 (B)	3317	1211 (35.5%)	5.53	29	30	59	85	119
Seq-S2w100 (B+S)	1794	1041 (58.0%)	7.31	40	57	73	94	125

Note:

S1w100: one-sample analysis for ChIP-seq data, window length w=100bp.

S2w100: two-sample analysis for ChIP-seq data, window length w=100bp.

B: applying boundary refinement.

S: applying single strand filtering.

The choice of window size w=100 bp represents a tradeoff between sensitivity and specificity (see Methods). Methods for motif mapping are described in Supplementary Methods online. A likelihood ratio LR≥500 was used as the cutoff to define NRSF motif sites. To facilitate a fair comparison between different datasets, the TRANSFAC⁴² NRSF motif M00256 was used in the motif mapping. Using the NRSF motif recovered from de novo motif discovery did not change the results qualitatively (data not shown).