Figure 1.

Histone modifications. Segtools plots for a segmentation of histone modification peaks produced with flatten (4: H3K4me3; 27: H3K27me3; 36: H3K36me3; 4/27: H3K4me3+H3K27me3; 4/36: H3K4me3+H3K36me3; 27/36: H3K27me3+H3K36me3; all: H3K4me3+H3K27me3+H3K36me3). A) The relative enrichment of these labels around active GENCODE release 3c protein-coding genes. "Manual" and "Auto" gene annotations from the UCSC Table Browser were merged and only protein-coding transcripts active in K562 (RPKM values in top 25%) were retained. Genes are split into idealized components: flanking regions; initial, internal, and terminal exons and introns, with the mean length of each component in parentheses. Enrichment is calculated as log2 (fobs + 1)/(frand + 1), where fobs is the frequency at which the given label is observed at the given offset, and frand is the frequency expected at random, given the relative abundance of each segment label. B) Mean dinucleotide frequencies across segments of each label.

Buske et al. BMC Bioinformatics 2011 12:415   doi:10.1186/1471-2105-12-415
Download authors' original image