Skip to main content


Fig. 3 | Genome Biology

Fig. 3

From: Clustered CTCF binding is an evolutionary mechanism to maintain topologically associating domains

Fig. 3

Representation of TE classes and their association with CTCF binding sites differs between TAD boundaries and other genomic regions. a Fractions of TAD boundary-associated versus non-TAD boundary-associated CTCF binding sites that are embedded in different TE classes. LINE-embedded CTCF sites are under-represented at TAD boundaries (χ2 test without Yates correction: p = 3.12e−15), while DNA transposon-embedded CTCF sites are over-represented (χ2 test: p = 0.0003), although accounting for just 3% of the TAD boundary-associated sites. SINE-derived CTCF sites (χ2 test: p = 0.01) and LTR-associated CTCF sites (χ2 test: p = 0.015) show no significant differences between the two categories. The top bar shows the percentage of the C57BL/6J genome sequence that corresponds to each TE class, for reference. b Fraction of sequence length of TAD boundary regions (TAD boundary ± 50 kb) occupied by each TE class, compared to random genomic regions of equal length. SINE sequences are significantly over-represented (Mann-Whitney U test: p < 2.2e−16), while LINEs are significantly depleted at TAD boundaries (p < 2.2e−16). DNA transposons are slightly, but significantly, enriched at TAD borders (p = 9.72e−14), although they account for only 1% of the sequences of the studied regions on average. Representation of LTR sequences shows no significant difference between TAD boundaries and random genomic regions (p = 0.005; significance threshold, 0.001)

Back to article page