Feasibility of using Oxford Nanopore Technology for population genomics in Fagus sylvatica

We conducted a pilot study to evaluate the feasibility of using Oxford Nanopore Technology (ONT) to analyze genetic variation among cohorts of European beech individuals in the context of a long-term monitoring program. This is desirable due to guarantees by ONT to support reference chemistry for long-term studies and the potential to recover structural and epigenetic variants as well as single nucleotide polymorphisms (SNPs). However, population genomics requires scalable approaches. We extracted DNA from dormant leaf buds of 18 adult trees and sequenced these either with individual barcodes or as a single pool with the aim of saving costs (approximately equimolar pooling) in duplicate (two replicate flow cells) using a PromethION. Results were generally reproducible across replicate flow cells but we observed strong bias in the number of reads across individually barcoded samples, likely due to differences in the intactness of the extracted DNA and small differences in size distribution that may have been amplified or further biased during library preparation. Furthermore, a disproportionately low number of sequencing reads was obtained for the mixed pool sample. Flow cell occupation by the 19 samples (18 individuals and mixed pool) resulted in a read depth that was too low to reproducibly detect structural variants or epigenetic markers, so to take advantage of the capabilities of long-read sequencing, higher read depth corresponding to fewer samples per flow cell and increasing cost would be required. Although seemingly a cost-saving measure, we generally do not recommend sample pooling as equal representation requires detailed, often cost-intensive characterization of samples prior to pooling. This makes low-coverage sequencing of individual samples the more feasible option in terms of cost-effectiveness. While low-coverage sequencing yielded interpretable SNP variation (which, however, remains more accurate using short-read sequencing techniques), it failed to yield reproducible data on epigenetic and structural variants. In conclusion, scaling ONT to population genomics would benefit from the development of more cost-effective methods to ensure even representation across samples and to obtain more reproducible information on variants, especially structural and epigenetic variants, with lower sequencing coverage.