plot a methylation graph from bismark's methylation data

Question

[alessiogentile1922]

Good morning to everyone, I'm a student from Italy. In my epigenetics course I' had an assignment where I had to align reads obtained from BS seq of an arabidopsis thaliana toy genome, align the reads to a reference and do a methylation call using Bismark. I'd obtained a methylome data called "CX_context" that is like this:

https://pastebin.com/BN248NXY

(1.6 gb txt file). Using this data from CX_context, my teacher told me to plot a graph of the methylation level along the chromosome like this:

https://imgur.com/a/IV506es

The hints that he gave to me were to use R studio, and search for an package that can do the "smoothing " on my data.

The question is... How can I check the methylation level if the methylation is a binary variable (methylated or unmethylated)? There's a package/tool that you can suggest me to use?

To do all these analysis we connetct via SSH to a linux based server.

Thanks to everyone that could help me.

Answer 1

[Charles Warden]

I think percent methylation is easier to visualize (although you may want to double-check the scale, since the absolute values are also relevant), and you want some way to make sure you aren't over-fitting your data (sometimes, it may be better not to smooth your data).

If this is BS-Seq data, there are a couple options that I can think of:

1) IGV has BS-Seq visualization. While you aren't directly looking at percent methylation, it colors the CpG sites differently and gives you a way to visualize percent methylation along with total coverage.

2) I think the Bismark quantification changes every so often. However, last time I checked, you could calculate the percent methylation from the counts table created using the following command:

$Bismark_path/bismark_methylation_extractor -o $outputfolder --buffer_size 10G --comprehensive --bedGraph --counts --no_overlap --report --genome_folder $ref -p $bismarkBam