NCM 2021: Evaluation of nanopore sequencing for epigenetic epidemiology: a comparison with DNA methylation microarrays

There has been increasing interest in the role of epigenetic modifications in the role of health and disease. DNA methylation can lead to both increases and decreases in gene expression. The main approach for profiling DNA methylation is chemical modification of unmethylated cytosine bases and PCR amplification, prior to sequencing. DNA methylation can be detected in the native DNA without the need for treatments using nanopore sequencing, which uses electrical signal profiles to distinguish between different nucleotide bases, including modified bases. It’s currently unknown whether this approach is sensitive enough to detect differentially methylated sites in an epidemiological study, or how the quantification compares to standard microarray technology. The Oxford Nanopore Cas9 targeted sequencing protocol was compared with microarray technology using previously identified methylated positions associated with tobacco smoking. Samples were taken from a heavy smoker and a non-smoker and sequencing was performed on MinION Flow Cells, generating 215,000 reads. An 8.3 kb region on chromosome 2 was sufficiently enriched using the nanopore protocol; however, the AHRR region on chromosome 5, which is 140 kb, required additional guide RNAs to ensure full coverage. Mean distance between guides was 15.6 kb. Profiling of the read length between guide RNAs suggested they are influenced by the size of the target region: the long reads only covered 10–20% of the region of interest. Nanopolish was used to quantify DNA methylation, which identified 1,779 CpG sites and nanopore technology was compared with microarrays using 98 of these sites: a strong positive correlation of 0.94 between platforms was observed. Eilis’ group then analysed the differences in DNA methylation between samples and compared this with microarray data. The nanopore data identified new differences, but not all of the previously identified sites were called, and this was found to be influenced by read depth. The nanopore data was concluded to be sensitive enough to study phenotypes associated with large effect sizes.