Extended Data Fig. 3: Analysis of 5mC and 5hmC from the same molecules revealed multiple 5hmC types.

a, Heat maps showing the 5mCHG and 5hmCHG TF motif enrichments along with TF expression level during mESC 2i state to serum state transition. b, Dotplots showing promoter methylation levels for genes associated with cell proliferation during mESC 2i to serum state transition. c, Cytosine modification entropies in single cells. Each dot represents a single cell and is colored and ordered according to its pseudotime score. Grey dots are background entropy levels estimated by shuffled the cell barcodes of called modification sites. d, Violin plots showing the fraction of 5mC and 5hmC reads with paired modality in single cells. For both Violin plots, hinges were drawn from the 25th to 75th percentiles, with the middle line denoting the median, whiskers with maximum 2× IQR. For 5mC modality, minima = 1.94%, maxima = 9.12%; for 5hmC modality, minima = 6.93%, maxima = 20.75%; n = 300 randomly sampled cells. e, Barplot showing the distributions of 5hmCG sites with different fractions of cells with detected 5mCG-associated 5hmCG sites. False positive detection rates (FDR) based on the fraction of averaged false detected (Type 2) sites from shuffled groups in total detected sites (FDR = 0.0467 was selected). f, UMAP embedding showing 5hmC levels of different types in single cells. Each dot represents a single cell and is colored according to the average 5hmC level. g, Top enriched GREAT GO terms for different types of 5hmCG sites. P-value, one-sided Fisher’s exact test. h, Histograms and heatmaps showing the relationship between two types of 5hmCG with mESC DNase-seq signals from ENCODE (ENCSR000CMW). i, Fraction of Type 1 and Type 2 5hmCGs in detected from low-entropy (0–75%) or high-entropy (75%-100%) cells. j, Barplot showing the enrichment of 5hmCG sites detected in low-entropy or high-entropy cells over different genomic regions.