Extended Data Fig. 5: Correction of standard (CaVEMan-based) mutation burden estimates and validation of NanoSeq indel calling.

a, Comparison of the mutation burden estimates in regions of the genome with at least 20× coverage (c) to the trinucleotide-context-corrected mutation burdens in the subset of c covered by NanoSeq and passing all NanoSeq filters. b, Ratio between the rates shown in a, showing that the corrected burden is approximately 20% higher than the uncorrected burden; box plots show the interquartile range, median and 95% confidence interval for the median. c, Comparison of indel rates between cord blood colonies (indels were called with the Pindel algorithm) and granulocytes from neonates (NanoSeq pipeline), showing Poisson 95% confidence intervals. Given the sparsity of indel calls in cord blood, data from different colonies (n = 100) and granulocytes (n = 2 donors, one of them with 5 replicates) were combined into single point estimates. d, The top two panels show the high similarity between the NanoSeq and Pindel indel profiles for a bladder tumour; the bottom two profiles show the indel spectra in blood from POLE and POLD1 germline mutation carriers, which are very similar to previously reported profiles⁴⁸.