Fig. 2

HMRs cluster more than expected. A Distribution plots of inter-HMR distances by cell type. The green distributions represent observed values from HMR datasets per cell type. Vertical navy bars show median values. Grey distributions show expected values by random shuffling across the non-coding genome. For each cell type, the expected and observed distributions were determined to be significantly different by the Wilcoxon rank sum test. All p-values were reported as zero (p < 2e-16) with a range of Χ² values from 1.4337 × 10⁸–4.4508 × 10⁸. B Diagram of HMR clustering and cell specificity workflow. HMRs are annotated for clustering behavior and/or cell specificity. Non-coding HMR datasets are defined by HMRs that do not overlap RefSeq protein-coding TSSs (TSS -2000/ + 1000) and exons. Clustering refers to groups of HMRs in a cell type that are located a maximum of 6 kb end-to-end from the next HMR, linking 3 or more HMRs; clusters cannot cross TSSs or exons. Unclustered HMRs are defined as non-coding HMRs that are not within 6 kb of any other non-coding or TSS/exon-overlapping HMR. Cell specificity is also defined, with any base pair overlap between HMRs constituting overlap. C Bar graph of HMR clustering annotations discussed in (B) and Fig. S6 as percentages of total HMRs by cell type. Selected cell types represent members of the hematopoietic and hepatic lineages. Colors reflect cell types representing different developmental stages and lineages. D Bar graph of proportion of cell type HMRs that are clustered HMRs (3 + HMRs) vs unclustered. Total values are calculated as [#unclustered + #clustered]. E Sankey diagram showing the flow of B cell HMRs. B cell HMRs are divided on the right of the panel into clustering groups. The left shows HSPC HMRs that overlap B cell HMRs, and are hierarchically categorized as clustered HSPC HMR, unclustered HSPC HMR, shared, or cell specific. To define cell specificity, B cell HMRs were compared to datasets from adrenal gland, H1 ESC, HSPC, fetal spinal, fetal heart, liver, macrophage, neutrophil, and T cell. F The bar graph shows the top biological process gene ontology results for the Sankey group of HMRs that progress from HSPC unclustered to B cell clustered (indicated in red). Results from GREAT Gene Ontology using default background and gene assignment settings are represented by bars showing binomial q-value [37]