How to use ENCODE data to find histone marks overlapping with 450k data

Question

[spurthy111]

Hi,

I am a first year PhD student and I have started working with 450k data recently.

I would like to use ENCODE data to check the overlap of all the available histone marks with our significantly methylated 450k array probes. We are trying to find which histone marks are most prevalent at those DNA sites where we see methylation differences. I was wondering if there are any tools available to do this. I have used ggbio, Annotationhub, but I was able to check only few types of histone markers. I appreciate any suggestions or tips.

Thanks for your time.

Regards,
Sailalitha.

Answer 1

[Alex Reynolds]

Split the histone mark signals into n separate, sorted BED files (mark_1.bed, mark_2.bed, ..., mark_i.bed, ..., mark_n.bed).

Set up another sorted BED file containing probe sites where there are methylation differences (diffs.bed).

Run a bedmap session to list the differential methylation site alongside the number of marks that overlap that site:

$ bedmap --echo --count diffs.bed mark_i.bed > answer_i.bed

You'll have {1..n} result files. Because all result files have the same sites, you could render all of their counts together in one figure, where the x-axis is sites and the y-axis is the count values at each site, and line plots are made for each mark. Line plots would quickly show where there is enrichment for one or more marks.

Alternatively, make a heatmap where rows are methylation sites and columns are mark counts. A tool like matrix2png is useful here. You could use row normalization to show relative mark enrichment at a site.

If you want to apply another statistic, such as one based on the mark signal value, look into the score operators, in place of --count.