Inference of 5hmC levels from BS-seq and oxBS-seq data

Question

[Mireia]

Hi,

I have sent samples WT and KO to be analyzed by BS-seq and oxBS-seq. I will receive BAM files, so I would like to analyse them with R software in order to infer 5hmC levels in each sample and compare.

I have seen 'adjustMethylC' function from 'methylKit' package. This function needs 5hmC levels from the same sample as input in order to adjust 5mC levels, so there should be a way to get 5-hmC levels.

Anyone know of any R package to get 5hmC levels from the subtraction of BS-seq and oxBS-seq results?

Answer 1

[Matthias Zepper]

You do not need a specific R package for that, you can just compare the coordinates of the methylation calls and the hydroxymethylation calls. The setdiff() function from the GenomicRanges package is most likely what you are looking for.

Since GenomicRanges are the standard for representing genomic locations within the Bioconductor project, I would recommend taking the time to familiarize yourself properly with those functions, as they will be insanely helpful for conducting any genomic analysis in R, e.g. mapping the calls later to genomic annotations.

[dariober]

I don't entirely agree with this answer... At each cytosine in the genome, (ox)BS-seq gives the count of Cs and Ts so you don't actually have positions that are methylated and positions that are not methylated. Therefore you cannot go for a simple intersection of the coordinates. For practical reasons you could, but you should be aware of the simplifications you are making.

[Matthias Zepper]

Then I might have misunderstood the question. I interpreted get 5hmC levels from the subtraction of BS-seq and oxBS-seq results as overlay the methylated and hydroxymethylated sites and return the ones that are exclusively hydroxymethylated in our dataset. Indeed, for differential methylation calling, there are more sophisticated approaches like hummingbird.

@Mireia: You will find all packages relating to methylation analysis in R in Biocondutor. There is a plethora of packages intended for RRBS/WGBS data, e.g methylKit, BiSeq or methylPipe and the respective vignettes will help you to select the one that best fullfills your needs.