Systematic Evaluation of Short-Read, Long-Read, and Microarray Platforms for Integrated Genomic and Epigenomic Profiling in Brain Tumors

Abstract Background Recent advances in sequencing- and microarray-based technologies for genome and epigenome profiling have created an urgent need to assess their analytical performance, technological maturity, and potential clinical utility. We therefore evaluated short-read 5-base sequencing, long-read Oxford Nanopore Technologies (ONT) sequencing, and EPICv2 microarrays using brain tumor samples as a clinically relevant benchmark. Methods The study included 5-base sequencing and EPICv2 microarray data generated for 16 brain tumor samples initially diagnosed as glioblastoma (GBM), together with ONT sequencing data available for 10 of 16 samples. Data were processed using technology-specific workflows, including DRAGEN somatic, EPI2ME wf-somatic-variation, and sesame R package for 5-base, ONT and microarray data respectively. The platforms were compared in terms of sequencing yield, basecalling quality, coverage of CpG sites and known problematic genomic regions, DNA methylation quantification, detection of genomic alterations in GBM-associated genes, and finally applicability to haplotype reconstruction. Results 5-base sequencing provided flexible sequencing yield, highest basecalling quality and CpG coverage, supporting scalable and accurate genomic and epigenomic profiling, whereas ONT sequencing allowed for uniform coverage of problematic genomic regions and variants phasing. Both sequencing platforms showed high concordance with EPICv2 methylation measurements, with Pearson correlation coefficient 95% CI of 0.95–0.96 for 5-base and 0.94–0.95 for ONT. Measurement discrepancies were greater at intermediately methylated CpG sites and generally decreased with increasing sequencing depth. In genomic analyses, 5-base sequencing, due to substantially higher yield-to-cost ratio, outperformed ONT in somatic small variant detection. At the same time, ONT long reads enabled uniform sequencing of problematic genomic regions as well as effective haplotype reconstruction and detection of allele-specific MGMT promoter methylation. Finally, integration of 5-base and ONT data further improved phasing performance. Conclusions The evaluated sequencing platforms provide complementary capabilities for genomic and epigenomic profiling. 5-base offers sequencing scalability, high basecalling accuracy, and broad methylome coverage, whereas ONT improves interrogation of problematic genomic regions and enables haplotype-resolved methylation analysis. EPICv2 microarrays provide cost-effective, scalable and highly consistent methylation measurements but remain restricted to predefined CpG loci.