Fig. 2

Characterization of ChIP-seq data. a Snapshots for AR (green), H3K27ac (orange), H3K4me3 (pink), and H3K27me3 (purple) ChIP-seq are shown at four example loci in four patients. Genomic coordinates are indicated above. b Peak width distribution of ChIP-seq peaks for AR (green), H3K27ac (orange), and H3K4me3 (pink) peaks. c Distribution of peak width for H3K27me3 ChIP-seq peaks. d Scores plot of principal component analysis based on occupancy (called peaks) of AR (green), H3K27ac (orange), H3K4me3 (pink), and H3K27me3 (purple) ChIP-seq samples. e Correlation heatmap based on peak occupancy. The clustering of the samples represents correlations between individual ChIP-seq samples on the basis of all called peaks. The column color bar indicates the ChIPped factor. Pearson correlation is plotted in white-green color scale. f Plot depicts the number of peaks overlapping in tumors for each factor ChIPped. Consensus peakset were chosen by using a cutoff of peaks present in at least 25, 40, 25, or 15 samples for AR, H3K27ac, H3K4me3, and H3K27me3, respectively. The number of consensus peaks is indicated for each factor. g Genomic distribution of consensus peaks from AR, H3K27me3, H3K27ac, and H3K4me3 across genomic features. h Bar chart shows the Z-score of the top 5 sequence motifs found at consensus AR peaks