Fig. 2

Comparisons of transcription profiles of three model breast cancer lines (MCF7, BT-549, BT-474) and a control non-cancer line (MCF10A). a Jensen-Shannon Divergence (JSD) values were calculated as the similarity of the probability distributions of expression levels (FPKM values) for 63,286 total transcripts, which include 22,267 protein-coding transcripts. In the lower grid, cummeRbund [66] was used to consolidate replicates and to calculate overall JSD between cell types. Solid border, BT-549 vs. MCF7, smallest JSD; dashed border, JSD’s for MCF10A vs. cancer cell lines. b The boxplots show gene expression values (center line, median; lower and upper boxes, 25th and 75th percentiles; lower and upper whiskers, minimum and maximum) for all protein-coding transcripts (22,267), H3K27me3-positive (1146) or EZH2-positive (2397) protein-coding loci. NS, no signal. c The Venn diagram includes HGNC symbols of genes that are H3K27me3-positive (middle box plot, b) and are silenced (FPKM < 2) in at least one cell type. GO term enrichment p-values are shown only for subsets where FDR < 0.1