Analysis of DNA 5-methylcytosine Using Nanopore Sequencing

Background

DNA methylation is one of the most important epigenetic modifications. It is involvedin the regulation of gene expression, gene imprinting, transposon silencing, andchromatin packaging in response to developmental and environmental stimulation [1].In mammals, DNA methylation mainly occurs in the CG context, and non-CG methylation(CHG and CHH) is only found in specific cell types such as brain cells andpluripotent cells [2]. In plants, however, in addition to the CG methylation, CHGand CHH methylations are also widespread throughout the genome and play importantroles in gene silencing [3]. There are various kinds of strategies for the detectionof DNA methylation. Among them, Illumina-based whole genome bisulfite sequencing(WGBS) can obtain global patterns at single-base resolution and thus serves as thegold standard for genome-wide DNA methylation analysis [4]. This method usesbisulfite to convert the cytosine into uracil; the latter turns into thymine afterPCR amplification, while the cytosine with 5mC can resist the conversion. Theirdifference can be identified by mapping to the reference genome after sequencing.However, WGBS has shortcomings like low coverage on repetitive regions due to theshort-read sequencing and false positives caused by incomplete conversion. Nanoporesequencing provides good solutions for all these problems. It allows the directdetection of methylation status on native single-molecule DNA without chemicaltreatment and PCR amplification. By studying the electric signal characters (called squiggle) produced when DNA goes through the Nanopore, the DNA sequence andassociated modification can be decoded [5]. The Nanopore long reads can cover largegenomic regions and enable the profiling of repetitive and complex regions as wellas the phasing haplotypes [6]. Currently, Nanopore sequencing can detect not only5mC but also other modifications like 5hmC, 4mC, and 6mA on DNA [7], and has beenapplied in bacteria [8,9], humans [10], and plants [11].

Many algorithms have been developed to decode the modification signal from Nanoporedata [7,12]. However, these methods failed to capture 5mC in the context of CHG andCHH with acceptable accuracies, which hinders their application in plant genomeresearch. To answer this requirement, DeepSignal-plant was developed. It uses deeplearning to recall 5mC in all contexts, having gained a high correlation with WGBSresults [13]. In this protocol, we introduce the data analysis process ofDeepSignal-plant for methylation study in plants.