Extended Data Fig. 2: Comparison of UBS-seq and conventional BS-seq using mESC gDNA.

a) With a cutoff ≥10× coverage and >10% of unconverted ratio, the number of the 5mC sites on non-CpG motifs detected by conventional BS-seq was twice as many as that detected by UBS-seq. b) The methylation levels detected in conventional BS-seq exhibited a slight increase at CpG sites in the 10 ng samples when compared to those in UBS-seq. c) Greater increases were observed in non-CpG sites. d) The sequencing coverage relative to a 1 kb sliding window was plotted against the GC% of the reference sequence in this region. The results indicate that conventional BS-seq exhibits greater coverage bias compared to UBS-seq. e) The unconverted ratio on non-CpG sites (CHG/GHH) show a slight decrease of unconverted ratio in regions with higher GC content for the both 10 ng and 1 ng UBS-seq libraries and the 10 ng conventional BS-seq library, while the 1 ng conventional BS-seq library shows an increase in the unconverted ratio in higher GC content regions. Comparison of the 10 ng and 1 ng samples of conventional BS-seq indicates that the conversion ratio at high GC regions of the lower input ratio is relatively higher. f) Comparison of genomic DNA coverage of DNA libraries prepared by conventional BS-seq and UBS-seq starting from 10 ng and 1 ng mESC gDNA, respectively. The average sequencing depth of sites within each 10 kb window was calculated and scaled to Z-score using the overall genomic coverage and plotted on a logarithmic scale. Outliers, indicated by black dots, were defined as values with an absolute Z-score greater than 1.5.