Poor small chromosome coverage returns error

[mlweilert]

This package will return subscripting errors when considering chromosomes that contain reads on only one strand (fwd versus rev) and not the other. This is a particular problem when looking at small, unannotated contigs in the dm6 and mm10 genomes.

We circumvent this by importing the BAM files with readGAlignments and then using the reads option in ExoData. Is there a way to fix this so we can use ExoData directly from the filepath?

[welch16]

Thanks for your feedback, I am aware of this issue, but I am not completely sure about how to fix it. On the top of my head I think that adding a chromosome argument, and then create the ExoData object is created for those chromosomes would fix the issue. Do you know in advance which chromosomes don't have poor coverage?

[mlweilert]

Forked your repo and created an automated fix that works really well for our lab. Made a pull request of the changes to your repo for your review in case you want to adopt it.

In this fix, we simply filter chromosomes that do not have at least one read from the fwd and rev strands aligned to that chromosome. This almost always happens in genome assemblies with chromosome contig portions like "chrUn_GL456392" in the dm6 genome assembly.

Creating a chromosome argument would also work, though!