Fig. 4: IT-scATAC-seq dissects cellular heterogeneity in human PBMCs.

a UMAP plots showing IT-scATAC-seq profiles of PBMC samples from healthy donors (IT, n = 7628), and two additional PBMC scATAC-seq datasets from healthy donors profiled by 10X (10X, n = 9411) and s3 (s3, n = 2855), coloured by sample origin. b UMAP visualisation coloured by cell type identity, including B cells, NK cells, T cells, monocytes and progenitors with a top panel showing cell type fractions for each sample. c Lineage-specific markers overlaid on the UMAP embedding, including PAX5, MS4A1, and EBF1 for B cells; CD3G, IL7R, and CD8A for T cells; FCGR3A (CD16), NKG7, and IL2RB for NK cells; and CD14, CEBPB, and CCR2 for monocytes. Visualisation is coloured by normalised gene scores, with the MAGIC algorithm used to smooth drop-out noise. d Heatmap showing Z-scores of normalised chromatin accessibility for 64,606 cell identity-specific marker peaks (FDR ≤ 0.1, log₂ fold change > 1) identified across scATAC-seq clusters using the two-sided Wilcoxon rank-sum test, with P-values adjusted for multiple comparisons using the Benjamini–Hochberg correction. e Heatmap displaying the top 10 TF binding motifs with the highest variability in respective marker peaks of each cluster ranked by adjusted P-value, calculated using a two-sided hypergeometric test, and P-values adjusted for multiple comparisons using the Benjamini–Hochberg method. f Scattered plot illustrating the correlation between motif accessibility and gene expression. Each point represents a TF, with the x-axis showing the correlation to gene expression and the y-axis indicating the maximum TF motif delta (variability) across clusters. P-values were derived from a two-sided Pearson correlation test and adjusted for multiple comparisons using the Bonferroni method. TFs identified as positive regulators (correlation >0.5 and adjusted P-value < 0.01, with max delta in the top quartile) are highlighted in red and other TFs in grey. Source data are provided as a Source Data file.