Comprehensive benchmarking of tools for nanopore-based detection of DNA methylation

Abstract

Abstract Oxford Nanopore (ONT) sequencing offers direct detection of DNA base modifications. Numerous tools have been developed to leverage this advantage. However, their performance remains unclear. Here, using diverse bacterial, plant, and mammalian datasets, we systematically evaluate the current landscape of nanopore methylation tools. We demonstrate that although most recent tools perform well, older models remain the reliable choice for studying CpG methylation. Conversely, newer models show substantial improvement in identifying 5-methylcytosine in non-CpG contexts, 6-methyladenine, and 4-methylcytosine. Further, we highlight the sensitivity of tools to confounding methylation nearby, assess their computational performance, and evaluate the effects of sequencing depth, methylation abundance, read quality, and basecalling mode. We provide reusable pipelines and open access datasets to empower future benchmarking efforts. Our work thus details the strengths and limitations of the state-of-the-art methylation models and outlines practical guidelines for researchers using nanopore sequencing to study DNA methylation.