Supplementary Fig. 2: Characterizing the chromatin interactions in serum- and 2i-ESCs. | Nature Cell Biology

Supplementary Fig. 2: Characterizing the chromatin interactions in serum- and 2i-ESCs.

From: Epigenetic modulation of a hardwired 3D chromatin landscape in two naive states of pluripotency

Supplementary Fig. 2

a) The percentage of differential chromatin interactions (y-axis) detected per category using different fold change cutoffs (x-axis). Decreasing the fold-change cutoff does not significantly increase the number of differential interactions for non-H3K27me3-associated loci. b) Insulated neighborhoods are highly similar between serum- and 2i-ESCs. The cumulative sum of all cHiC interaction-reads covered within an insulated neighborhood is shown. N number indicated in the panel represents insulated neighborhoods. R-values represent Pearson’s correlation coefficient. Differential insulated neighborhood was computed using two sided Wald tests (DEseq2) and adjusted for multiple comparison using Benjamini-Hochberg correction. c) The insulation scores of strong and week CTCF loops are similar between serum- and 2i-ESCs. p=2.1e-107 (ser) and p =8.5e−113 (2i), Wilcoxon rank-sum test (two-sided). n=11,441 strong-strong and n=5,222 weak-weak boundaries. Box=5-75th percentile; bar=median; whiskers=5-95th percentile. d) The sum of interaction reads of all connected enhancers per promoter is highly similar between serum- and 2i-ESCs with the exception of H3K27me3-associated promoters. Circle size: number of interactions per promoter. n=250 up-regulated and n=250 down-regulated genes, R-values represent Pearson’s correlation coefficient. e) Scatter plots showing the intensities of promoter interactions for the 250 most induced genes in serum- and 2i-ESCs. Even after increasing the CHICAGO-score and read count cutoff, the vast majority of non-H3K27me3-associated genes show similar interactions in serum- and 2i-ESCs. Genes with differential interactions (FDR < 0.1) mainly encompass genes decorated by H3K27me3 (shown in blue). f) Grid analysis of the number of significant interactions found by increasing the interaction read count (x-axis) and CHICAGO significance threshold (y-axis). While the number of interactions rapidly decrease (panel 1, blue), the number of differential interactions remains low, but may increase due to a lower multiple correction penalty (panel 2, green). Increasing the read and score cutoff resulted in a minor increase in the number of differential non-H3K27me3 interactions (panel 3, yellow). Increasing the CHICAGO significance threshold biased towards long-range interactions, dominated by CTCF-CTCF loops (panel 4, red). Increasing the read count cutoff biased the analysis towards short-range interactions, which have a higher proximity ligation background. g) Scatter plots showing the intensities of promoter interactions for the 250 most induced genes in serum- and 2i-ESCs across different biological replicates. Similar interaction intensity in both culture conditions is observed even after increasing the sequencing depth five-fold. R-values represent Pearson’s correlation coefficient. n=250 genes.

Back to article page