Fig. 1

Identification of partially hypermethylated UMRs. A The computational framework for identification of hypermethylated regions from normal and cancer methylomes. B Bimodal distribution of the proportion of hypermethylated CpGs sites on under-methylated regions in IDH-mutant glioma. C Genome browser visualization of representative phUMR (SYT6) and fhUMR (TPPP3) in IDH-mutant gliomas. refUMR represents consistently under-methylated region in multiple brain tissues. Bottom panel represents methylation level track in a larger genome scale and the corresponding expression signals. D Comparison of methylation changes calculated using the hypermethylated regions (y axis) versus the entire reference under-methylated regions (x axis). Methylation change of each region was the absolute difference of mean methylation level for all CpG sites between IDH-mutant gliomas and normal brain samples. The frequently used methylation difference threshold 0.2 is denoted as the red line. E Comparison of methylation levels of SYT6 and TPPP3 using different benchmarks (n = 75 for normal brain tissues and n = 15 for IDH mutant glioma tissues). Hyper, hypermethylated regions in phUMRs or fhUMRs