Profiling genome-wide methylation in two maples: fine-scale approaches
to detection with nanopore technology
Abstract
DNA methylation is critical to the regulation of transposable elements
and gene expression and can play an important role in the adaptation of
stress response mechanisms in plants. Traditional methods of methylation
quantification rely on bisulfite conversion that can compromise
accuracy. Recent advances in long-read sequencing technologies allow for
methylation detection in real time. The associated algorithms that
interpret these modifications have evolved from strictly statistical
approaches to Hidden Markov Models and, recently, deep learning
approaches. Much of the existing software focuses on methylation in the
CG context, but methylation in other contexts is important to quantify,
as it is extensively leveraged in plants. Here, we present methylation
profiles for two maple species across the full range of 5mC sequence
contexts using Oxford Nanopore Technologies (ONT) long-reads. Hybrid and
reference-guided assemblies were generated for two new Acer
accessions: Acer negundo (65x ONT and 111X Illumina) and
Acer saccharum (93x ONT and 148X Illumina). The ONT reads
generated for these assemblies were re-basecalled, and methylation
detection was conducted in a custom pipeline with the published
Acer references (PacBio assemblies) and hybrid assemblies
reported herein to generate four epigenomes. Examination of the
transposable element landscape revealed the dominance of LTR
Copia elements and patterns of methylation associated with different
classes of TEs. Methylation distributions were examined at high
resolution across gene and repeat density and described within the
broader angiosperm context, and more narrowly in the context of gene
family dynamics and candidate nutrient stress genes.