Calculation of analysable genome size

To estimate the somatic mutation rate, it was first necessary to establish the size of the analysable nuclear genome (that is, the portion of the genome in which variant calling could be performed reliably) for each sample (Supplementary Table 4). For both substitutions and indels, the analysable genome of a sample was defined as the complement of the union of the following genomic regions: regions reported as ‘not analysed’ by the CaVEMan variant caller; regions failing the ‘chromosome and contig’ filter; regions failing the ‘N-tract and contig-end’ filter; and regions failing the ‘sequencing coverage’ filter (see ‘Variant filtering’). For the analysis of mitochondrial variants, the analysable genome of a sample was defined as the portion of mtDNA that satisfied the ‘sequencing coverage’ filter (see ‘Mitochondrial variant calling and filtering’), after subtracting the hypervariable region (D-loop).