Figure 2.

Whole-genome RE1 screening

(A) Contents of the whole-genome RE1 library. ChIP-seq peaks from 4 cell types with or without a canonical REST-binding motif were combined with 5 enhancers. The names of enhancers correspond to the benchmarking set shown in Figure 1. The motif shown is from JASPAR (MA0138.2).

(B) Proportion of RE1 that showed repression with 1% of false discovery rate (FDR) in at least one cell/enhancer combination.

(C) Log₂(odds ratio) for enrichment of epigenetic marks in the strong silencers shown in (B). ∗∗adjp < 0.01, ∗∗∗adjp < 0.001 by Fisher’s exact test corrected using BH.

(D) Correlation between predicted binding score of REST and motif contribution (log₂(scrambled/native)) with En19 in K562. The purple line indicates linear regression, and the orange line indicates piecewise linear regression. The dashed line indicates the change point determined by the piecewise linear regression.

(E) MPRAduo activity with En19 for each cell type. Genomic REST ChIP+ indicates RE1s bound by REST in each cell type and genomic REST ChIP− indicates RE1s not bound by REST in the observing cell type but bound in other cell type(s). Gray line indicates the median of the negative control. ∗adjp < 0.05, ∗∗adjp < 0.01, ∗∗∗adjp < 0.001 by U-test compared with negative controls corrected using BH.