We analyzed a recently published single-cell DNA accessibility and mitochondrial genome sequence co-assay dataset from a patient with a colorectal cancer (CRC) tumor16. We first performed QC, dimension reduction and clustering on the DNA accessibility assay and annotated the major cell types present in the dataset based on the DNA accessibility at key marker genes (Fig. 4a). This revealed five major clusters present in the dataset encompassing tumor-derived epithelial cells, basophils, myeloid cells and T cells, as previously reported16. To identify clonal relationships between cells in the CRC dataset, we identified highly variable mitochondrial genome positions among the cells by computing the variance-to-mean ratio and the Pearson correlation between strand coverage (Fig. 4b). Visualization of per-cell allele frequencies (fraction heteroplasmy) for these variants in the two-dimensional UMAP space computed using the DNA accessibility assay revealed the variant 16147C>T present at nearly 100% frequency in tumor-derived epithelial cells, whereas other variants were shared across different immune cell types (Fig. 4c). We further identified cell clones by clustering allele frequency data, revealing ten distinct clones (Fig. 4d). Clones 1, 2 and 4
