Fig. 2

Effect of substitutions and temperature on bias for homogeneously methylated templates. The rows represent the different substitutions at the variable sites. C primers matched the methylated template sequence (C in the forward primer, G in the reverse), I primers had inosines, Y primers had a C/T degeneracy (A/G in the reverse), N primers had totally degenerate bases (A/T/C/G), mm primers had mismatches (A in the forward primer and T in the reverse primer) and ab primers had abasic sites. The columns represent the annealing temperatures tested. For each assay, the average methylation percentage as measured at each CpG by bisulfite pyrosequencing across the amplicon is shown. The final CpG in both assays was not used in the calculations of the average methylation across the amplicon as the pyrosequencing measurement was consistently lower than expected for a homogeneously methylated template and was called as inaccurate by the software. The annealing temperature increases along the x-axis by 2 °C for each step as indicated. The mismatched (mm) primers for CDKN2B used different annealing temperatures as indicated but were still used in 2 °C increments. The colour scales were set so that the target measurement reflecting the actual level of methylation is shown as white (25 in the 25% methylated templates (panel a) and 50 in the 50% methylated templates (panel b)). Because of this, the same methylation values have different colours for the 25 and 50% dilutions. The scales at the bottom represent the colour coding for the 25 and 50% mixtures, respectively. The numbering in each scale indicates the lowest value for each colour, i.e. 11 represents 11–20