Resolution of the DNA methylation state of single CpG dyads using in silico strand annealing and WGBS data

Whole-genome bisulfite sequencing (WGBS) has been widely used to quantify cytosine DNA methylation frequency in an expanding array of cell and tissue types. Because of the denaturing conditions used, this method ultimately leads to the measurement of methylation frequencies at single cytosines. Hence, the methylation frequency of CpG dyads (two complementary CG dinucleotides) can be only indirectly inferred by overlaying the methylation frequency of two cytosines measured independently. Furthermore, hemi-methylated CpGs (hemiCpGs) have not been previously analyzed in WGBS studies. We recently developed in silico strand annealing (iSA), a bioinformatics method applicable to WGBS data, to resolve the methylation status of CpG dyads into unmethylated, hemi-methylated, and methylated. HemiCpGs account for 4-20% of the DNA methylome in different cell types, and some can be inherited across cell divisions, suggesting a role as a stable epigenetic mark. Therefore, it is important to resolve hemiCpGs from fully methylated CpGs in WGBS studies. This protocol describes step-by-step commands to accomplish this task, including dividing alignments by strand, pairing alignments between strands, and extracting single-fragment methylation calls. The versatility of iSA enables its application downstream of other WGBS-related methods such as nasBS-seq (nascent DNA bisulfite sequencing), ChIP-BS-seq (ChIP followed by bisulfite sequencing), TAB-seq, oxBS-seq, and fCAB-seq. iSA is also tunable for analyzing the methylation status of cytosines in any sequence context. We exemplify this flexibility by uncovering the single-fragment non-CpG methylome. This protocol provides enough details for users with little experience in bioinformatic analysis and takes 2-7 h.