Fig. 2

Properties of mutations in B-N-P-T tetra samples. a M over m preference in LOHs. LOH-M/LOH-m ratios for LOHs arising at germline Mm positions are shown for N-, P-, and T-stages (data from Fig. 1b, right panel). b Dinucleotides in genomic sequences captured by AluScan (blue bars) and SNVs found at the first base of dinucleotides (red bars). c Percentile of different types of sequential SNV and CNV changes in 12 tetra sample cases of BRCA, STAD, and LIHC. Type I (green) B=N, viz. no SNV (or CNV) found in ∆NB. Type II (yellow) B ≠ N = T, viz. same SNV (or CNV) found in ∆NB and ∆TB. Type III (blue) B ≠ N ≠ T and B ≠ T, viz. altered in ∆NB and ∆TN and also in ∆TB. Type IVa (red) B ≠ N ≠ P = T and B = T, viz. altered in ∆NB and ∆PN but not in ∆TP or ∆TB. Type IVb (purple) B ≠ N=P ≠ T and B = T, viz. altered in ∆NB and ∆TP but not in ∆PN or ∆TB. d Average distances between GOHs (left) or LOHs (right) and their nearest recurrent CNVs (rCNVs). The left panel shows that on average, CG>TG GOHs in ∆NB were closer to their nearest rCNVs than it was the case with 11 other kinds of C>T GOHs; the right panel shows that on average, TG>CG LOHs in ∆PN were closer to their nearest rCNVs than it was the case with 11 other kinds of LOHs, with the horizontal bars indicating how much closer in terms of p values. e Correlation between patient’s age at diagnosis and percentage of CG>TG GOHs among all GOHs for ∆NB (left) or percentage of TG>CG LOHs among all LOHs for ∆PN (right). f Correlation of numbers of somatic CNV breakpoints found in tumors (CNVT) from the COSMIC database with either evolutionarily conserved CpG-rich regions (CpGe) (left panel) or unmethylated CpG-rich regions (MeMRE) (right panel) from UCSC Table Browser database [35]. p, significance probability; r, Pearson correlation coefficient (see Fig. 1 for abbreviations)