plotEnrichment or a similar tool

[dpryan79]

It's been internally requested that we have a tool that can produce things like (A) FRiP scores or (B) histograms of reads per feature type (exons, UTR, etc.). Such a tool might be called plotEnrichment. The output would be something like an image and an optional table of values. The input would be a BED or GTF file. One difficulty is that deeptoolsintervals is currently only storing a single feature type. libGTF, on which it's based, can store everything, so this shouldn't be a real problem. What I expect I'll do is implement a slightly tweaked GTF/BED parser only for this tool that will end up using the same C code. Then, we'll use mapReduce (possibly in a slightly modified form) to chop the genome into bins and then iterate over reads in each (after first checking to ensure that there's at least one GTF/BED feature in said region).

I guess there's a reason that I implemented counting overlap sets in libGTF...

[dpryan79]

There's now a feature/plotEnrichment branch where this is largely implemented (though it's a definite work in progress) via the command plotEnrichment. The image output is as below:

peaks are any regions in a BED file, whereas for GTF files the feature column is used. I would suggest that if people want introns that they annotated them (or just vaguely estimate them as the difference between gene and exon, though the real value will be higher).

[dpryan79]

@thomasmanke, since you were asking about this.

[dpryan79]

As requested, BED files now get individual labels (either the file name or whatever you specify with --regionLabels). GTF files will always use the feature column and ignore --regionLabels. Here are two bed files (genes19.bed and genesX.bed) with new labels:

[dpryan79]

I've merged the branch containing this into develop. I'll add the Galaxy wrapper on Monday and then close the issue.

[dpryan79]

The wrapper is now included (and being tested by planemo).