AGBT 2025

Robert Osborne1, Fabio Puddu1, Annelie Johansson1, Aurélie Modat1, Jamie Scotcher1, Riccha Sethi1, Shirong Yu1,2, Nick Harding1, Mark Hill1, Ermira Lleshi1, Casper Lumby1, Jean Teyssandier1, Michael Wilson1,3, Robert Crawford1, Tom Charlesworth1, Shankar Balasubramanian1,4,5, Páidí Creed1
1biomodal Ltd, The Trinity Building, Chesterford Research Park, Cambridge, UK.
2Current address: Tagomics Ltd, The Cori Building, Little Abington, Cambridge, UK.
3Current address: Princeton University, New Jersey, US.
4Cancer Research UK Cambridge Institute, University of Cambridge, Cambridge, UK.
5Yusuf Hamied Department of Chemistry, University of Cambridge, Cambridge, UK.
Early cancer detection has the potential to significantly improve treatment outcomes and survival rates. This study investigates the roles of 5-methylcytosine (5mC) and 5-hydroxymethylcytosine (5hmC) as biomarkers for early-stage colorectal cancer (CRC) detection in cell-free DNA (cfDNA). Using whole genome sequencing, we analyzed cfDNA from 37 treatment-naive CRC patients and 32 healthy controls. Our findings indicate that combining measurements of 5mC and 5hmC significantly enhances diagnostic accuracy (AUC = 0.95) compared to traditional approaches that conflate these markers (modC). Notably, 71.7% of differentially methylated regions (DMRs) exhibiting an increase in 5hmC in stage I cfDNA also showed a corresponding decrease in 5mC in stage IV, suggesting that 5hmC can effectively track regions of demethylation during tumor development. These results support the hypothesis that distinguishing between 5mC and 5hmC can improve the sensitivity of liquid biopsy tests for early cancer detection.

Kenneth Beckman, John Garbe, Nick Jones, Mark Murphy, Noah Strom 

University of Minnesota Genomics Center (UMGC), Minneapolis, MN, USA 

The genome-wide analysis of DNA methylation at CpG (5mC) sites can be achieved by several methods, with array-based approaches from Illumina having dominated the analytical space for two decades, and sequencing-based approaches (WGBS, RRBS) becoming common as NGS costs have dropped. The nature of the information provided by DNAm arrays is strikingly different from NGS. Compared to arrays, which interrogate tens of thousands of molecules and provide CpG site beta values (5mC/C) with high precision, NGS typically interrogates tens of reads at a given locus, with single-site 5mC/C precision limited by read depth. At the same time, compared to NGS, which provides allelic information on neighboring CpG sites found within a single read and reports on all mappable CpG sites in the genome, arrays mask allelic information and target a small fraction of CpG sites. Until recently, moreover, neither arrays nor NGS have allowed for the facile genome-wide measurement of hydroxymethylcytosine (5hmc) along with 5mC.  

With the introduction of the duet multiomics solution evoC from biomodal, genome-wide 6-base sequencing (A, C, T, G, 5mC, 5hmC) is possible in a single workflow. Consequently, we have benchmarked biomodal’s evoC to Illumina’s Infinium HumanMethylationEPIC v2 and MethylationScreeningArray-48 in order to provide bi-directional insights into the platforms. We find that in terms of CpG site beta values, the platforms are well correlated (with quantitative agreement limited, as expected, by sequencing depth), bolstering confidence in both. Our comparison reveals that arrays are less able to distinguish extremes of methylation state (100% methylated or 0% methylated) than evoC and suggests that by leveraging allelic information from neighboring CpG sites, the precision of lower depth sequencing to deliver a comparable beta value to arrays may be improved. We have modeled the use of evoC to complement arrays in terms of performance, cost, and utility in epigenetic clock analysis, and have attempted to generate a list of array CpG site reliability using evoC data as a truth model.