Extended Data Fig. 8: Multiplexed scRRBS integration with single-cell transcriptomes and genotyping.

a, Schematic of the joint multiplexed scRRBS, transcriptome and genotyping capture protocol. b, Normalized Robinson–Foulds distances between any two trees (n = 30 tree replicates; see Methods) of CLL12 (n = 56 cells; see Fig. 3h) reconstructed by maximum-likelihood versus maximum-parsimony analyses. Differences (Δ) are indicated. c, Proportion of wild-type (white) and mutated (black) SF3B1 cells in each clade identified from the lineage tree shown in Fig. 3h. d, Comparison of the number of unique CpGs (left) and the CpG methylation level (right) between the wild-type-enriched and the mutated-enriched SF3B1 clade of cells identified from the lineage tree in Fig. 3h. e, Volcano plot of differentially methylated gene promoters (absolute weighted average DNAme difference > 0.3 and two-sided non-parametric permutation test P < 0.05) between the wild-type and mutated SF3B1 cells from the lineage tree shown in Fig. 3h. f, Single-cell alternative 3′ splicing score (fraction of reads that map downstream to the 3′ end (up to 100 bp) of the exons versus within the exons) for cells belonging to wild-type (n = 30) and mutated (n = 26) SF3B1 clades identified from the lineage tree shown in Fig. 3h. g, Volcano plot of differentially expressed genes between the wild-type-enriched and mutated-enriched SF3B1 clade. Genes (n = 57) with absolute log₂(SF3B1 mutated-enriched/SF3B1 wild-type-enriched gene expression) > 0.5 and Benjamini–Hochberg FDR-adjusted weighted F-test P < 0.2 are shown in red. Genes that were previously reported to be affected by SF3B1 mutation²² are also labelled. h, Gene expression projections on lineage trees for two representative genes identified in g. i, Comparison of transcriptional distances (measured as Euclidean distances of the first three principal components after principal component analysis) as a function of lineage distance between cell pairs from the lineage tree shown in Fig. 3h. j, Cells belonging to SF3B1-mutated enriched clade show significantly lower relative node heights (that is, height of internal tree nodes relative to the root node; see Methods) compared with wild-type SF3B1-enriched clade, consistent with SF3B1 mutation being a later subclonal event in CLL¹⁵. k, As in j for root-to-tip branch lengths (that is, the length from the root to each tip in the lineage tree). l, Distribution of node ages (estimated number of divisions before present; see Methods) between the wild-type (white, n = 30 nodes) and mutated (grey, n = 25 nodes) SF3B1 enriched clade. Box plots are as defined in Fig. 1. P values were determined by two-sided Mann–Whitney U-test (b, d, f, i, j, l) or two-sided Fisher’s exact test (c).