Fig. 3: Canonical cancer-associated genes with CGI methylation differences and concordant expression involving germline SV breakpoints.

a Venn diagrams representing overlapping top CGI probes between the CBTN-based germline SV-methylation-expression associations (from Fig. 2c) and probes for genes with established cancer association by COSMIC⁴⁴ (left) or for known cancer susceptibility genes from the literature⁴⁵. Left diagram is for genes positively or negatively associated with SV breakpoints, with corresponding CGI methylation association in the opposite direction; right diagram, for negatively associated genes with positive CGI methylation association. Significance of overlap by one-sided Fisher’s exact test. b Left: CGI methylation (purple) and mRNA (orange) levels of MSH2, corresponding to SV breakpoints in the genomic region surrounding the gene. Each point represents a patient. Methylation and mRNA values are respectively normalized to standard deviations from the median (using logit- and log2-transformed values, respectively). Right: Boxplots of MSH2 CGI methylation (beta) and log2 expression by germline SV class. P-values versus other by t-test on logit-transformed (methylation) or log2-transformed values (RNA). c Similar to part b, but for RPSA-associated germline SVs and RPSA differential CGI methylation and expression. d Similar to part b, but for PALB2-associated germline SVs and PALB2 differential CGI methylation and expression (p-values compare DGV:esv1921182³⁴ SV, or “esv. SV”, versus the rest of patients). DGV:nsv4236915 (n = 4 patients), or “nsv. SV”, in the same PALB2 downstream region, was not associated with differential methylation. e Across 202 impacted patients, MSH2 germline SV events versus somatic mutation events involving MSH2, H3-5, H3-3A, and TP53 (left), along with their respective associations with overall tumor somatic mutation burden (right); p-values by t-test. All boxplots represent 5%, 25%, 50%, 75%, and 95%, with n representing the numbers of patients.