Fig. 5

Effect of biases and artefacts on the output of 5mC quantification. a Average methylation values of iDMRs. Numbers indicate the mean value, error bars span the 10–90 percentile. b Genome-wide comparison of absolute methylation levels for the amplification-free PBAT approach and the Heat+KAPA BS-seq. c Differences in the absolute quantification of genomic and regulatory features between the amplification-free PBAT (at value 0) and amplified WGBS datasets. Numbers indicate the mean value, error bars show the 10–90 percentile. d Comparison of relative methylation differences between sperm and ESC sequenced with either amplification-free PBAT or Heat BS-seq. Each dot represents a probe over 150 consecutive cytosines from the same genomic region in ESC and sperm. The plotted over 20% mCG differences are generated from the BS-seq method (left panel) and visualised with the same colour onto the PBAT data (right panel). Averaged values were used for BS-seq (2 × ESC and 5 × sperm replicates) and a single replicate for PBAT. e Venn diagrams showing how many of the over 20% mCG regions from d overlap between BS-seq and PBAT. f A breakdown of relative contribution of biases for the BS-seq protocols as measured by LC-MS and WGBS. For post-bisulfite protocols, the overall combined bias is shown as individual contributions are less trivial to dissect. The non-BS 5mC measurement averages LC-MS measurements for mESC lines from different studies and passages to account for culturing and lineage variances. Error bars represent standard deviation