Skip to main content
[Preprint]. 2023 Apr 4:2023.04.02.535219. [Version 1] doi: 10.1101/2023.04.02.535219

Figure 3. Predicting cCREs in the VISION project and comparisons with other catalogs.

Figure 3.

(A) Method for calling cCREs using S3V2-IDEAS in the IS mode. The normalized ATAC-seq signals (expressed as the negative log10 p-value for fitting a negative binomial distribution, signal range 0–10, 200bp bins) are shown for a selected subset of the 39 human biosamples plus the average signal track in an 11kb genomic interval around the transcription start site (TSS) of the ITGB2B gene is shown (GRCh38 chr17:44,384,001–44,395,000). The signal intensity states learned by IDEAS in the IS mode are shown as shades of violet (state 0 is white, darker shades represent higher signal states). Genomic intervals in high signal states were called as peaks (yellow rectangles) in a four-step hierarchical process designed to limit the peak calls to local maxima while also finding cell type-specific peaks (see Methods). Peaks in this genomic region illustrate calls at steps 1, 2, and 4 of the hierarchical process. Panels B-F present intersections of human VISION cCREs with genome-wide datasets of structural and CRE-related elements. (B) The Venn diagram displays the result of intersecting VISION cCREs with a combined superset of elements associated with nuclear structure (CTCF OSs, loop anchors, and boundaries) and with a combined superset of DNA intervals associated with cis-regulatory elements (CREs). (C) The proportions of cCREs and randomly selected, matched sets of intervals in the overlap categories are compared in the bar graph (right). For the random sets, the bar shows the mean, and the dots show the values for each of ten random sets. (D) A higher resolution view (an UpSet plot) of the intersections of VISION cCREs with the four groups of CRE-related elements, specifically enhancer-related (Enh), transcription start sites (TSS), Survey of Regulatory Elements (SuRE), and CpG islands (CpG). The enrichment for the cCRE overlaps compared to those in randomly selected, matched sets of intervals are shown in the boxplots below each overlap subset, with dots for the enrichment relative to individual random sets. (E) Overlaps and enrichments of VISION cCREs for three sets of structure-related elements, specifically CTCF OSs (CT), loop anchors (LA), and TAD boundary elements. (F) Overlaps of VISION cCREs with two sets of experimentally determined blood cell cCREs.