Fig. 3
Relationships between genetic variants and genome instability. a Average recombination intensity in 10 equal groups of 1-kb sequence windows ranked according to SNP density from low (left) to high (right). R2 is the coefficient of determination. b Higher recombination intensity in 1-kb windows with GVs compared to windows without GVs (unpaired one-tailed t tests, Bonferroni corrected). c Enrichment of eight kinds of genetic-variant hotspots, recombination hotspots (Rec-Hs) and somatic CNVTs in the six DNA replication phases, expressed as density fold-changes relative to autosomal average. d Densities of GWAS-identified SNPs (corrected for allele-frequency dependency, see “Statistical analysis” in “Methods” section), breakpoints of variants in ClinVar database, and somatic CNVT breakpoints in total hotspots and total clusters relative to randomly simulated genomic regions represented by the gray violin-plots. All p values smaller than 0.05 are shown in red. e Recurrency of somatic CNVT breakpoints in eight kinds of hotspots, in “Hotspot” representing all 44,379 genetic-variant hotspots in the genome, or in “Cluster” representing all 1135 hotspot clusters in the genome, with their p values estimated by the Monte Carlo method. f Percentage of hotspot- or cluster-containing genes found in Tumor Suppressor Gene Database (TSG) and Network of Cancer Genes (NCG). The p values estimated using chi-square tests are shown above each bar (Bonferroni corrected). In (b, d e, f), arrow on y-axis indicates the autosomal average. Error bars in (b), and shaded bands around the curve in (a) indicate 95% confidence intervals. “H” stands for “hotspots”