Single-cell mitochondrial variant data processed using mgatk16 was downloaded from Zenodo (https://zenodo.org/record/3977808), read into R using the Signac function ReadMGATK and used to create a Seurat assay. Informative mitochondrial variants were identified using the IdentifyVariants function, which computes the strand concordance in variant counts (Pearson correlation) and the variance-mean ratio (VMR) for each variant, as previously described16. Informative mitochondrial variants were selected with a VMR > 0.01 and strand concordance ≥0.65, provided the variant was confidently detected in ≥5 cells. We then computed per-cell mitochondrial allele frequencies for informative variants using the AlleleFreq function in Signac.
