Figure 3.

Functional analysis of primary tumor and relapse sample

(A) Number of CpGs in regions found to be differentially methylated in the sample comparison (primary tumor vs relapse) as well as ASM in the two samples. Colors represent an estimation of discoverability with short-read sequencing methods. CpGs in low-complexity regions (soft-masked in reference) are more difficult to map using only short reads. CpGs not phaseable with short reads are further than 150 bp from a phased heterozygous non C>T variants.

(B) Methylation of NRN1 promoter and enhancer in the primary tumor sample.

(C) Heterozygous deletion in promoter of PTCH1 (tumor-suppressor gene and driver in medulloblastoma) with differential methylation in the remaining haplotype.

(D) Predicted gene fusion pairs from Arriba validated using ONT long-read information, thresholded by confidence as reported by Arriba. Fusion pairs in the “supported by individual reads” category are supported by at least one genomic read with a chimeric alignment including both genes. Pairs in the “explainable using genomic breakpoints” category have a plausible explanation by following a graph of structural variations that connect the two genes. The category “high confidence read support” refers to pairs where both these criteria are met.

(E) Example of a gene fusion pair that can be explained using genomic breakpoints but with no individual genomic read that covers both genes. Two separate insertions of a total length of 42,797 bp appear to be involved in the fusion of LINC01091 and FKBP9 such that, even in ONT reads, there was no read extending across the entire gene fusion.

(F) PCDH17 (tumor-suppressor gene) promoter with ASM pattern in the primary tumor sample.