Fig. 3

Adjacent segment analysis. a Adjacent segment pairs. Row: code of the 5′-segment, one row per code. Left: Distance in number of modification differences of the epigenetic code on H1. Columns represent the distances (0–3) between the code of the 3′-segment (column) and the code of the 5′-segment (row). Middle: Frequency of adjacent segments with specific code combinations. Columns represent the code of the 3′-segment. Right: Observed-vs-expected frequencies of mark combinations of adjacent segments. Expected frequencies are calculated based on the code frequencies in the data sets. Columns represent the code of the 3′-segment. Blue colors indicate combinations that are observed less often than expected while red colors indicate combinations that are observed more often than expected. Stronger saturation implies larger distance from equality. Black indicates combinations not observed at all. b Over-represented peak arrangements: (Case 1) More often than expected by chance, H3K4me3 peaks are located within an H3K27me3 domain. Less frequent but still more often than expected by chance, H3K27me3 peaks are located within an H3K4me3 peak. (Case 2) We also observe Case 1 being located within a large H3K9me3 domain more often than we would expect by chance