Characterization of background errors in targeted deep sequencing data. a The background allele frequencies from plasma and PBL DNA samples were box-plotted (n = 19 for each group). b The frequency of error-free positions in each sample was calculated and box-plotted for each group. c The distribution of background allele frequencies across all possible 12 substitution classes. The y-axis denotes the frequency of each class in the pre-treatment PBL and plasma samples. The relative base substitution frequency is shown in the stacked bar plot on the right side. d The ratio of background errors in PBL compared to plasma DNA samples was plotted for each substitution class indicated on the x-axis.
e The ratio of background allele frequencies for reciprocal base substitutions. Error bars indicate standard deviation