Mapping DNA methylation with high-throughput nanopore sequencing

Abstract

DNA chemical modifications regulate genomic function. We present a framework for mapping cytosine and adenosine methylation with the Oxford Nanopore Technologies MinION using this nanopore sequencer's ionic current signal. We map three cytosine variants and two adenine variants. The results show that our model is sensitive enough to detect changes in genomic DNA methylation levels as a function of growth phase in Escherichia coli.

Figure 1

Cytosine methylation variant calling accuracy results on synthetic oligonucleotides. Results are from classification of 6,966, 294, and 467, C, 5-mC, and 5-hmC strands respectively that were barcoded and sequenced in the same MinION flow cell. A. Per-read accuracy distribution is shown for the maximum-likelihood estimate (MLE) normal distributions and the ‘Multiset’ HDP model. The triangles represent the mean of the distribution. B. Average three-way classification accuracy for all sites on the substrate. Dotted lines represent the mean across all sites for template (blue) and complement reads (green). C. Confusion matrix showing HMM-HDP three-way cytosine classification performance on template reads of synthetic oligonucleotides. D. Scatter plot showing the correlation between the log-odds of correct classification and the mean pairwise Hellinger distance between the methylation statuses of the 6-mer distributions overlapping a cytosine.

Figure 2

Observed and learned ionic current distributions and read accuracy correlation with ungapped alignment score for 6-mA in GATC motifs (left) and 5-mC in CC(A/T)GG motifs (right). Top: Comparison of the influence of 6-mA and 5-mC on ionic current levels for representative 5-mers. The empirical ionic current levels from 100 aligned events are shown as a normalized histogram and the HDP-learned probability densities were are shown as curves. The HDP density was sampled on 900 point grid from 50 to 140 pA. Bottom. Correlation between ungapped alignment score (see Methods) and per-read accuracy for 500 randomly sampled template reads.

Figure 3

Changes in genome-wide cytosine methylation at different stages of culture growth. Bar height represents the percentage of residues that were called as methylated. Axes are broken to accentuate differences between the growth phases.