Bacterial DNA methylation amplifies the shared A-to-G and G-to-A substitution bias in nanopore sequencing

Abstract Nanopore sequencing enables direct analysis of native DNA modifications, but these modifications can also introduce systematic basecalling errors. Here, we investigated how bacterial DNA methylation contributes to nanopore substitution errors using matched native whole-genome sequencing (WGS) and methylation-depleted whole-genome amplification (WGA) datasets from six bacterial species across CycloneSEQ (CS) and Oxford Nanopore Technologies platforms (ONT). Adenine-to-guanine (A2G) and guanine-to-adenine (G2A) substitutions were consistently enriched across species, nanopore platforms, sequencing configurations, and basecalling models, indicating a shared nanopore-associated substitution bias. This enrichment remained detectable in WGA reads, whereas native WGS reads showed a further increase. Using independent PacBio methylation profiling, we found that the WGS-associated substitution excess was strongly concentrated around bacterial methylation motifs and was minimal in non-motif regions. The magnitude and substitution pattern of these errors varied among motifs and sequencing platforms. At a subset of motif-associated positions, alternative-base fractions exceeded 50%, producing SNP-like signals. In ONT data, 28 WGS-specific calls survived the applied SNP-calling filters, all within PacBio-supported methylation contexts, and most showed significant strand imbalance. Together, these results identify a shared A2G/G2A substitution bias in all nanopore datasets and support a methylation-associated increase in these errors near bacterial methylation motifs, with some loci producing SNP-like artifacts.