Fig. 2: Read-level methylome classification results. | Nature Communications

Fig. 2: Read-level methylome classification results.

From: MethylBERT enables read-level DNA methylation pattern identification and tumour deconvolution using a Transformer-based model

Fig. 2

In all line plots, the error bars indicate 95% confidence interval. A–C An example of simulated read-level methylation patterns for tumour and normal cell types (reads were simulated with the complexity of a1_b5, see Methods for details) and the classification accuracy comparison across different complexities (A) with read length 150 bps, (B) with read length 500 bps, and (C) with CpG-specific methylation patterns. Yellow and black on each read show methylated and unmethylated CpGs, whereas grey represents other bases. Sequence reads for two cell types are divided by a dotted line in the middle. Region-wise methylation levels in tumour and normal cell types are shown in the histogram. D Read-level methylome classification accuracy comparison for different read coverages. E MethylBERT read-level methylome classification accuracy for different read coverages and pattern complexities.

Back to article page