Abstract
BACKGROUND: The detection of 5-methylcytosine (5mC) patterns in the human genome is relevant for the diagnosis of various genetic conditions. Genome-wide methylation episignatures provide a new approach for resolving variants of uncertain significance. Whole-genome DNA methylation detection is usually performed using short-read bisulfite sequencing procedures or methylation arrays. Pacific Biosciences (PacBio) long-read sequencing (LRS) is a technology that can detect 5mC on native DNA. Here we assessed the ability of PacBio HiFi LRS to robustly identify genome-wide methylation for medical purposes. We assessed the ability to identify differentially methylated imprinted regions and genome-wide epigenetic signatures for genetic disorders. METHODS: PacBio LRS methylation detection was compared to two conventional genome-wide DNA methylation sequencing techniques. We used 30 PacBio LRS samples (10 trios at ~ 30× coverage) and assessed the ability to detect differential methylation using 25 known parentally imprinted regions with different genome-wide sequence coverage. Finally, we evaluated two published epigenetic signatures for KMT2A gene defects using PacBio LRS data from 12 KMT2A patients, 2 patients with a variant of uncertain significance, and 39 control samples. RESULTS: PacBio methylation calls on the HG002 cell line were highly comparable to short-read protocols, providing 5% additional calls, predominantly in regions that are difficult to assess using short-read technologies. Among the 10 trios, correct haplotype-resolved methylation patterns were found for 24 out of 25 (96%) imprinted regions and all of the methylated alleles showed the expected parental origin. Downsampling analysis showed that these imprinted regions can robustly be detected using a minimum of 15× genome-wide coverage. Finally, we found that although the two published KMT2A episignatures are significantly different, both are successfully able to distinguish KMT2A patients from controls and classify the two KMT2A VUS samples as control samples. CONCLUSIONS: Overall, our results indicate that PacBio LRS can reliably detect specific medically relevant methylation changes as well as genome-wide episignatures in rare disease patients.