Skip to main content
Fig. 3 | Genome Biology

Fig. 3

From: A probabilistic generative model for quantification of DNA modifications enables analysis of demethylation pathways

Fig. 3

Identifying differentially methylated cytosines. a A ternary plot representation of the posterior distribution of the methylation proportions (estimated over three replicates) for the cytosine at chr4:139,783,857 in v6.5 (left panel) and Tet2kd (right panel) samples. b A comparison of Lux, FET, and MOABS in detecting differential methylation. For this purpose we down-sampled the full data set to 12× coverage for each of the three replicates. BS-seq and oxBS-seq data sets were analyzed separately with FET and MOABS for differential methylation. All the covered cytosines in a CpG context (N = 384) were divided into sets of differentially (N = 252) and similarly (N = 132) methylated cytosines based on independent CMS-IP and MeDIP loci-level information (see "Materials and methods"). The ROC curves of the methods are calculated based on the differential methylation analysis results. The curves of different methods (Lux, FET, and MOABS) and data types (BS-seq/oxBS-seq) are distinguished with different colors. The AUC values are listed in the figure key. c Same as (b) but here the methods are compared using simulated data (see also Figure S8e in Additional file 4)

Back to article page