Predicting unrecognized enhancer-mediated genome topology by an ensemble machine learning model

(Downloading may take up to 30 seconds. If the slide opens in your browser, select File -> Save As to save it.)

Click on image to view larger version.

Figure 1.
Figure 1.

H3K27ac and YY1 HiChIP demarcate active enhancer loops. (A) Proportion of annotated loop types for K562-H3K27ac HiChIP data. Each loop identified with an FDR < 0.05 and pair-end tag number ≥ 2. ChromHMM was used to annotate the anchors, with only enhancer and promoter type anchors being retained. The majority of loops are enhancer-mediated (79.7%). (B) Super-enhancer plot for anchors in K562-H3K27ac HiChIP data. Slope threshold was set at 1000. A total of 880 super-enhancers were found, which accounted for 5.6% of all enhancers. Super-enhancer signal derived from H3K27ac ChIP-seq data. (C) GO analysis for super-enhancer anchors in K562-H3K27ac HiChIP data (binomial P-value < 1 × 10−10). (D) De novo motif enrichment analysis on K562-H3K27ac loops. Transcription factors ranked by normalized gene expression. The size of each point indicates the motif enrichment P-value. The transcription factors with high motif enrichment (–log2[P-value] > 100) and gene expression are shown. The YY1 motif was significantly enriched (−log2[P-value] = 306). (E) Diagram depicting the putative co-enrichment of H3K27ac and YY1. (F) Top: Venn diagram showing the intersect of K562-YY1 and K562-H3K27ac HiChIP data sets. Bottom: Proportion of each annotated loop category from the overlapping H3K37ac and YY1 HiChIP loops (n = 132,928). The majority of the overlapping loops were enhancer-associated (86.4%). (G) Genome browser tracks showing H3K27ac and YY1 ChIP-seq signals and topological interactions.

This Article

  1. Genome Res. 30: 1835-1845

Preprint Server