Extended Data Fig. 4: Analysis of 3D genome topology in ES cells expressing GFP-CTCF or the del8 mutant.
a) Principal component analysis (PCA) of the top 1000 genes with highest standard deviation in ES cells expressing del8 (d8) or GFP-CTCF (GC). We performed RNA-Seq on three biological replicate cultures of three independently isolated d8 lines (d8.1, d8.2, d8.3) and two GC lines (GC1, GC2). One RNA-Seq sample of clone d8.3 showed poor quality metrics and was therefore excluded. b) Reproducibility scores (based on spectral decomposition; on a 0 to 1 scale, see Methods for details) across in-situ Hi-C matrices between the indicated pairwise comparisons of conditions. Note high reproducibility scores (>0.7) amongst biological replicates, which significantly drop (<0.4) when correlating in-situ Hi-C datasets from GC and d8-expressing cells. c) Distance decay curves of chromosome-wide interactions for wildtype and del8 expressing ES cells. d) Absolute PC1 values of genome-wide A and B compartment regions (100 kb bins) that switch compartment when comparing GC and d8-expressing cells. Data shown are merged values from three independent biological replicates. Center lines denote median values; box limits indicate 25th-75th percentiles with whiskers extending 1.5 times the interquartile range. e) Change in mRNA expression levels as measured by RNA-Seq for genes located in genomic bins that switch compartment when comparing GC- and del8-expressing cells. Data shown are values from 2-3 independent biological replicates obtained from 2-3 independently generated ESC clones. Center lines denote median values; box limits indicate 25th-75th percentiles with whiskers extending 1.5 times the interquartile range. f) Absolute PC1 values of genome-wide A and B compartment regions (100 kb bins) for untreated and CTCF-depleted cells. Data were obtained from ref. 29. Data shown are merged values from two independent biological replicates. Center lines denote median values; box limits indicate 25th-75th percentiles with whiskers extending 1.5 times the interquartile range. Pie chart indicated genomic regions (100 kb bin size) that switch compartment (4.1% of the genome) due to CTCF depletion. ***P < 0.001; A: P = 0.00039, B: P = 0.00081 (two-sided Mann Whitney U test). g) 2D meta-plot quantifying inter-TAD interaction frequencies between superenhancers in untreated and CTCF-depleted cells (ref. 29).
