Extended Data Fig. 1: Sequential chemical labeling enables simultaneous detection of 5mC and 5hmC bases on the same molecules.

a, Overview of TAPS and hmC-CATCH. b, Sanger sequencing results showing the ‘C-to-T’ conversion signal from model oligonucleotide sequence (T5MH) before treatment, after 5hmC labeling, and after both 5hmC and 5mC labeling. c, Schematics of chemical labeling for 5hmC and 5mC. d, qPCR result of lambda DNA (25,086bp-25,308bp) before and after potassium ruthenate (K₂RuO₄) treatment; n = 3 (Treated), n = 3 (Control). Data are presented as 6.020 ± 0.006 (Treated) and 5.957 ± 0.026 (Control). e, Agarose gel images of dsDNA of fragmented lambda DNA treated with 5hmC-labelling reaction only (left panel) and sequential 5hmC and 5mC-labelling (right panel). Experiment was performed once. f, Barplot showing the distribution of C-to-T mutation rates for unmethylated CH sites and methylated CG sites. g, C-to-T mutation signals on both strands of T1M spike-in model DNA, symmetric methylated CG sites are indicated in red on the sequences below. h, Genome browser view showing the sequenced reads aligned to spike-in model DNA (upper: T1M with 5mCG, positive and negative strands are separately displayed, bottom left: T2H with a single 5hmCG site, bottom right: T3MH with both 5mC site and 5hmCG site at known position). C-to-T is colored in red for positive strands, and G-to-A is colored in green for negative strands, respectively. Conversion rates are estimated from all the modified cytosines of spike-in model DNA. i, Nuclei clumps and resolved single nuclei suspension after brief sonication under bright field microscope. Experiment was performed once. Scale bars, 100 μm. j, Enrichment analysis of genome coverage by SIMPLE-seq and DNase-seq on DHS (DNase I hypersensitive sites). k, A table showing tagmentation efficiency under different Tn5 reaction conditions, including Tn5 with hyperactive mutations, working concentrations and reaction buffers; right-sided showing the fragments analysis result under optimal condition. l-n, Standard curves for mixed oligo DNA with (l) 5mC and C, (m) 5hmC and C, (n) 5mC and 5hmC, were plotted based on gradient mixing ratios (0:10; 2:8; 4:6; 6:4; 8:2; 10:0). o, C-to-T mutation rate estimated from 10-kb non-overlap bins across the whole genome. For both boxplots, hinges were drawn from the 25th to 75th percentiles, with the middle line denoting the median, whiskers with maximum 2× interquartile range (IQR). For 5mC, minima = 1.12%, maxima = 1.47%; for 5hmC, minima = 0.75%, maxima = 2.43%. 5mC, n = 156,755; 5hmC, n = 159,962. p and q, Scatter plot showing (p) the number of 5hmC reads mapped to human and mouse genome and (q) the fraction of 5mC and 5hmC reads mapped to human and mouse genome in each cell from the species-mixing experiment with twin axes. r, Stacked barplot showing the fraction of reads mapped to the reference genome and assigned to 5mC, 5hmC, or cannot be assigned. s-v, Comparisons between SIMPLE-seq and published single-cell and bulk 5mC and 5hmC sequencing methods: (s) fraction of mappable reads, (t) the number of 5mCG sites detected and total sequenced reads in each study, and (u) the number of 5hmCG sites detected and total sequenced reads in each study, (v) Dot plot showing the average number of covered CGs and sequenced reads for each cell in each study. w, Line plots showing the number of unique mapped reads or CG dinucleotides with at different sequenced read depths per cell in each study.