Integrating CpG-level methylation and transcriptomics for high-resolution cancer epigenetics

Background: Promoter hypermethylation is a key mechanism for silencing tumor suppressor genes in cancer. While whole-transcriptome sequencing (WTS/RNA-seq) is widely used to study cancer biology, high-resolution integration of methylation and expression data remains challenging. Conventional differential methylation region (DMR) approaches aggregate signals across promoters, potentially obscuring fine-scale regulatory effects. Methods: We analyzed three pairs of parental and sotorasib-resistant NSCLC cell lines using PredicineEpic genome-wide DNA methylation profiling and PredicineWTS RNA-seq. Instead of promoter-level aggregation, we quantified fragment-level methylation changes at individual CpG sites within promoter regions. Correlations between CpG-specific methylation alterations and differential gene expression were evaluated across varying TPM log₂ fold-change thresholds. In addition, more than 50 FFPE tissue biopsies from multiple cancer types were profiled with PredicineEpic and PredicineWTS to systematically assess the relationship between DNA methylation and gene expression. Results: CpG-level methylation changes showed stronger inverse correlations with gene expression than promoter-level averages. For genes with log₂FC ≥ 4, CpG fragment beta differences correlated with expression changes at R = −0.92 (n = 7, p = 0.001). Similar trends were seen at lower thresholds: log₂FC ≥ 3 yielded R = −0.63 (n = 22, p = 0.001), whereas promoter-level beta differences showed no significant correlation. Consistent patterns of stronger CpG-level correlations were also observed across all tissue biopsy datasets. Conclusions: High-resolution CpG-level methylation analysis provides greater sensitivity for linking epigenetic alterations to transcriptional changes than conventional promoter-level approaches. Fragment-level methylation profiling can reveal critical transcriptional regulatory events and may have important applications in liquid biopsy and biomarker discovery.