Fig. 6. The chromatin landscape of human MBCs.

a Clustering of CD27^– NBCs and CD27⁺ total MBCs based on top differentially accessible loci as displayed by PCA (4198). Prediction ellipses define 95% confidence intervals. Each symbol represents an individually sorted subset (n = 4). b Overlap of accessible loci between NBCs and total MBCs depicted by Venn diagram. c Differential chromatin accessibility (p < 0.05) between total MBCs and NBCs as determined by DEseq2. Differentially accessible regions (DARs) with increased (red) or decreased (blue) accessibility depicted by histograms. d Proportion of DE genes that overlap with DARs, as depicted by Venn diagrams. Percentage of genes exhibiting corresponding increases/decreases in chromatin accessibility shown below the number of genes. e Statistically over-represented KEGG pathways associated with increased (red) or decreased (blue) DARs depicted by the histogram. f Significant TF- binding motifs enriched in NBC- or all MBC-specific open chromatin regions as determined by HOMER motif analysis. g For all motifs (dots), the changes in MD score between NBCs and total MBCs (y-axis) are (MA) plotted against the number of motifs within 1.5 kb of any ATAC-Seq peak center (x axis)—computed by DAStk⁹⁴. Red points depict statistically significant increased MD-scores (p < 0.05). h, i RNA-Seq and ATAC-Seq data sets were integrated using the Taiji PageRank algorithm⁹² to generate PageRank activity scores and p-values for human TFs. h All 421 TFs (p < 0.05) plotted in rank order according to their log₂PageRank score ratio between MBCs vs. NBCs. Each TF PageRank −log₁₀(p) is denoted by color-coding according to legend: more significance, red; less significance, blue. i The top 10 TFs in NBCs and top 10 TFs in MBCs was plotted according to their log₂PageRank score ratio and log₂fold change in transcript expression in MBCs vs. NBCs. The bubble size of each TF is determined by the −log₁₀(p) of Taiji analysis.