Extended Data Fig. 4: Transcriptome and accessibility profiling in the PF1 cell line.

a, Number of unique reads mapped (in millions) per library for triplicate total RNA-seq replicates. b, Pairwise scatters of RNA measurements for all annotated mouse genes. Values are shown as log2(TPM + 0.01). Upper panels represent the Pearson correlation coefficient for the respective scatter. c, Same as in A but for unique reads in triplicate ATAC-seq samples. d, Pairwise scatter of reads in merged peaks across 3 ATAC-seq replicates. Axes represent log2(reads per kb + 1). e, Heatmap of ATAC-seq counts in a 10 kb window surrounding transcription start sites. Rows are ordered by TPM from RNA-seq data in (a), and represented as the annotation column to the left of the heatmap. f, Upper: stripchart of template (black points) and non-template (gray points) mutation rates divided by the total genic mutation rate for all 14 genomes. Point clusters represent genic bins as described in Fig. 4. From left to right, ROS mutations unique to each sister cell (14 points per bin), ROS mutations shared between sisters (7 points per bin), and UV mutations (14 points per bin). Lower: Boxplot of template - non-template rate for all 14 genomes, considering the mutations as for the stripchart panels above. Boxplot elements are as in Fig. 3c without notches. g, Average ATAC signal over gene bodies. Genes at least 5 kb in length were first binned based on TPM from low (1, light blue) to high (4, dark blue), and additionally 2500 coordinate shuffled gene positions (gray) were taken as a negative control. Gene bodies were divided into 100 tiles. Additionally, a window of 5 kb was added flanking the TSS and TTS. Reads were counted in all genic tiles, summed by genic bin, and scaled to reads per kb of genomic representation. h, Genic signal for Flag-OGG1 ChIP data in HEK293 cells⁴². Transcriptional binning and gene body tiling were performed as in panel (g), and numbers of genes per bin are shown as in (g).