Extended Data Fig. 5: Validation of ReDeeM lineage tracing via dual-tracer experiment.

a, The design of the dual lineage-tracer experiment with a Kras;Trp53(KP)-drive lung adenocarcinoma lineage-tracing mouse model. CRISPR-based and ReDeeM-based lineage information were analyzed for the same cells. 6 Independent tumours were profiled in batch1 (T1-T6), and 4 tumours were profiled in batch2 (T7-T10). b, CRISPR indel-based lineage relationships were computed with weighted hamming distance and visualized by multidimensional scaling (MDS). Tumour 1 (T1) is shown as an example with 214 single cells. c, Example mtDNA somatic mutations that agree with the CRISPR indel-based MDS map are highlighted. d, Schematic of the Agreement of Closeness (AOC). e, The phylogenetic trees based on mtDNA mutations are illustrated. The “clonal groups” are indicated as the colored bars. The positive AOC ratio for each clonal group is shown within each colored bar. The individual AOC scores (middle) and mutation numbers (bottom) are shown for every single cell (each leaf). The p values are computed by 1000 times permutations (one-sided, Supplementary Methods). The whole tree-level metrics of positive AOC ratios are shown for each tumor below the colored bar. f, Adjusted Rand Index (ARI) for clonal cluster consistency between CRISPR and ReDeeM across 10 tumors (T1-T10). Various clonal cluster resolutions are tested (presented as each dot). n = 16 resolution pairs for each tumor. Boxplot displays data from the 25th to 75th percentile, and whiskers extending to the minimum and maximum within 1.5 IQR. g-h, Illustration of the clonal cluster consistency for T1 (214 single cells) and T9 (410 single cells) on CRISPR-based embedding map using one example cluster resolution. Colors indicate either ReDeeM-based or CRISPR-based cluster identification. Also see Supplementary Methods, Supplementary Figs. 2–3.