See also Supplementary Fig. 1. a Overview of the study. HSC: hematopoietic stem cell, (pre-)LSC: (pre-)leukemic stem cell, Blast: mature leukemic blast. b Overview of the MutaSeq method. Targeting primers (purple) are included during the cDNA amplification step of the Smart-seq2 protocol. Targeting primers are directly fused to illumina library adapters (blue) and therefore get amplified efficiently during library preparation. Tagmentation introduces the same adapters to the full-length cDNA product. c Number of target sites covered per cell, across n = 206 (Smart-seq2) or n = 658 CD34+ (MutaSeq) bone marrow cells from patient P1 (see the Methods section Data visualization, for a definition of boxplot elements). d Mean gene expression of genes containing mutations of interest is plotted against the fraction of cells in which the mutation is covered. e Fractions of cells covering key non-synonymous mutations observed in the patient. Reference call: The reference allele was observed. Mutant call: the mutant allele, as defined by bulk exome sequencing (Supplementary Data 1) was observed. Mutant+Reference: both alleles were observed. f Allele frequency estimates derived from deep exome sequencing compared to allele frequency estimates derived from MutaSeq (red dots, n = 2208 single cells) or Smart-seq2 (gray dots, n = 206 single cells). Dot size indicates coverage at target site. Point shape indicates the type of mutation. Error bars indicates interquartile range. See figure source data for complete specification of sample size used to derive statistics (n). g Coverage of the mitochondrial genome in MutaSeq data. See Supplementary Fig. 1j for a comparison across methods. Green line segments correspond to genes in the mitochondrial genome.