Fig. 3: Transcriptional strand asymmetry of indels that occur at polyN tracts across multiple cancer types.

a Transcriptional strand asymmetries of indels occurring at polyT motifs. Average bias is shown with error bars showing standard deviation after 1000 bootstraps. Myeloid, cervix and thyroid cancers were excluded due to low numbers of total indels (Supplementary Table 1). T template, NT non-template. Strand bias was calculated as mutational density of non-template strand over total mutational density (of non-template and template strands). b Strand bias of MSI and MSS samples in stomach, biliary, uterus and colorectal tumours (Mann–Whitney U p value < 0.001 in all cases, Bonferroni corrected). c Transcriptional strand asymmetries of indels occurring at polyG motifs. Average bias is shown, with error bars showing standard deviation from bootstrapping. d Relationship between indel strand bias and gene expression levels in lung cancer (Mann–Whitney U p value < 0.001 for comparisons between low and medium expressed genes and between medium and highly expressed genes) according to length of polyG tracts (Kruskal–Wallis H-test with Bonferroni correction, p value < 0.001 for medium and high expression genes, p value > 0.05 for low expression genes). e Scheme depicting mechanism of indel mutagenesis at polyT tracts. DNA damage, shown as asterisks (*) that arise at T nucleotides of polyT tracts can occur on both template and non-template strands. The subsequent DNA repair, postulated to be TC-NER, results in preferential correction of DNA damage on the template strand, leaving T insertions (highlighted in as red T’s) and T deletions (shown as red −) on the non-template strand. f Schematic depicting mechanism of indel mutagenesis at polyG tracts in lung cancers from smokers. DNA damage in the form of adducted guanines (*) is asymmetrically repaired by TC-NER, with preferential repair of the template strand, thus accumulating more G indels on the non-template strand.