Table 1.

Distribution of NFI-C-occupied sites at H3K27me3 and H3K36me3 boundaries

29,158 H3K27me3 boundaries, subdivided in:	Matching NFI-occupied site	Matching random genomic sites	P-value (binomial)
4,687 H3K27me3 boundaries co-localized with H3K36 boundaries	148 (3.15%)	45 (0.96%)	1.51e-14
2,032 H3K27me3 and H3K36me3 boundaries co-localized with TSS	35 (1.72%)	6 (0.29%)	7.84e-07
1,507 H3K27me3 and H3K36me3 boundaries co-localized with TES	21 (1.39%)	11 (0.72%)	0.050
1,148 H3K27me3 and H3K36me3 boundaries not co-localized with TES or TSS	92 (8.01%)	28 (2.43%)	1.11e-09
24,471 H3K27me3-only boundaries (not co-localized with H3K36 boundaries)	284 (1.16%)	253 (1.03%)	0.167
3,386 H3K27me3-only boundaries co-localized with TSS	67 (0.19%)	19 (0.05%)	5.29e-08
2,153 H3K27me3-only boundaries co-localized with TES	20 (0.09%)	13 (0.06%)	0.162
18,932 H3K27me3-only boundaries not co-localized with TES or TSS	197 (1.04%)	221(1.16%)	0.889

29,158 H3K27me3 modification boundaries were mapped over the murine genome using the ChIP-Part algorithm, and this dataset was intersected with the H3K36me3 boundaries to yield co-localized or non-overlapping boundaries datasets. Boundaries were defined as co-localized if the sequences 2.5 kb up- and down-stream from each boundary overlap by at least 1bp. Colocalization of NFI with the H3K27me3 and H3K36me3 boundaries, or H3K27me3-only boundaries, was assessed using the set of 4,794 occupied sites or using the control set of 4,794 randomly selected genomic sequences. The percentages indicate the fraction of the particular dataset indicated in the left column that overlaps either an NFI site or a randomly selected site.