Quantitative detection of DNA methylation from nanopore sequencing data without raw signals

Abstract Background Nanopore sequencing has revolutionized the field of epigenomics by enabling direct detection of DNA methylation without additional sample preprocessing such as bisulfite treatment. It is theoretically possible to reutilize any nanopore sequencing data to construct epigenomes. However, reutilizing the data in practice is challenging with existing methods because they rely on raw signals from nanopore sequencing, which are absent in more than 98% of public nanopore sequencing data. Moreover, storing raw signals for large-scale sequencing projects is impractical due to their enormous file sizes. Findings To overcome these limitations, we propose a novel method, NanoFreeLunch, which can quantitatively detect DNA methylation without the need for raw signals by modeling base quality values and sequencing error patterns. Our results demonstrated a strong correlation between the DNA methylation levels estimated by NanoFreeLunch and those estimated by the benchmark methods, ranging from 0.87 to 0.94 for individual genomic loci and from 0.97 to 0.99 for average methylation levels of genomic regions. Conclusions NanoFreeLunch enables detection of DNA methylation from nanopore sequencing data without raw signals. With the rapid accumulation of nanopore sequencing data, the development of NanoFreeLunch will enable the construction of epigenomes on an unprecedented scale, facilitating novel insights into the role of DNA methylation in health and disease.