Fig. 4

Effect of conversion artefacts on the biases in WGBS. a Presence of unconverted cytosines as percentage of total cytosine content, measured by LC-MS for three different BS-conversion protocols. The three protocols differ by denaturation method (Heat or Alkaline) or molarity of bisulfite (4.5 vs 9 M for Am-BS) but not by BS incubation temperature (65–70 °C). Averaged fold differences in quantity are shown above horizontal brackets, and a dotted line shows the usual level of genomic 5mC for reference of scale. For conversion differences between methods with 50 and 65 °C incubation temperatures, see Additional file 2: Figure S10a. b A theoretical sum of 5mC and unconverted C as measured by LC-MS for J1 WT mESCs for three BS conversion protocols. Both 5mC and unconverted C will be interpreted as 5mC after amplification of WGBS libraries, boosting the overall levels of methylation, depending on the BS treatment protocol. c Absolute quantification of unconverted cytosines in the unmethylated TKO mESC line, as measured by Heat and Alkaline BS-seq. d Context distribution of BS conversion artefacts; the value is the same for Heat and Alkaline and therefore plotted as an average. e CH methylation on both strands of the mouse major satellite repeat as measured by pre- and post-bisulfite WGBS methods. 5mC percentage from the BS cloning from Additional file 2: Figure S5a is plotted in both panels for reference. Positive y-axis values indicate the top strand and negative the bottom strand. Statistical analyses in a–c were performed for matched experimental pairs with unpaired two-tailed t-tests against Heat in a and c, and WT ES in b. Error bars in a–c represent standard error of the mean, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001