Fig. 1

Schematic overview. Condition-specific Regulatory Units Prediction (CRUP) is a three-step framework to predict active enhancers (CRUP-EP), assign them to dynamic conditions (CRUP-ED), and create differential regulatory units (CRUP-ET). a CRUP-EP accounts for the size of accessible regions (highlighted in blue) which are flanked by nucleosomes. For each region of interest, bin _x, a combination of two binary random forest classifiers, solely based on ChIP-seq HM data, is then used for enhancer prediction. b Based on a permutation test, CRUP-ED computes empirical p values for each bin _x across different conditions (dotted and solid rectangles), which are further used to combine and cluster regions. c CRUP-ET inspects each differential enhancer region (blue ellipse) within its topologically associated domain (blue triangle). To infer putative target genes, the correlation between probability values and gene expression counts is calculated