Genomic visualization function

[rcavalcante]

Issues to resolve before implementation

A couple of questions come to mind:

• Should we support visualization of a BSseq object without differential methylation tests?
• Should we change the diff_ functions to include a flag like include_sample_meth which would be TRUE/FALSE and would append columns for sample-wise methylation to the diff_ function results?
  • There is value in doing that outside of plotting (we'll end up doing it in EpiCore pipelines anyway).
  • If we don't add an include_sample_meth flag, we would also need to require the BSseq object to append that data anyway.
• Is this the kind of plotting function that requires wrapping in png() ... dev.off()?

Function call

plot_methylation(bs, diff_result, ...)

Description

A function to plot the result of diff_binomial(), diff_methylsig(), or diff_dss_test() on a genome track. Users may pass options through this function to GViz::plotTracks() to visualize other tracks alongside the methylation data.

Arguments

• bs a BSseq object.
• diff_result a GRanges object obtained from diff_binomial(), diff_methylsig(), or diff_dss_test().
• ... parameters passed to GViz::plotTracks().

Values

An image (?) or an object (?).

Tests

If a file of the plot is returned, not sure. If an object, check correct class.

[rcavalcante]

All plots will have some constant components:

• GenomeAxisTrack()
• IdeogramTrack(genome=gen, chromosome=chr)
  • We'll need to either pass a genome build to use with GenomeInfoDb::Seqinfo(), or make sure to grab the seqinfo from the object we're plotting.
• GeneRegionTrack(geneModels, genome=gen, chromosome=chr, name="Gene Model")
• plotTracks(dTrack, groups=rep(c("control", "treated"), each=3), type=c("a", "p"), legend=TRUE) does grouping and has many plotting type choices here. For group methylation data, it feels like a boxplot grouped by case and control is most appropriate.
  • We can either include -log10(p-value) track, or it would be really nice to include some sort of significance identifier (, , , for example), like we often see in papers.