How to obtain low mappability regions from UCSC for deepTools computeGCBias?

Question

[chiefcat]

Hi!

Does anyone know how to download low mappability regions from UCSC for computeGCBias?

In deepTools documentation, it said, "You can download the mappability tracks for different read lengths from UCSC, e.g. for mouse and human."

When I go to UCSC https://hgdownload.cse.ucsc.edu/gbdb/mm10/bbi/, there are different files, which one should I use?

Or alternatively, can I just use the blacklists from ENCODE and modENCODE consortia instead? https://sites.google.com/site/anshulkundaje/projects/blacklists

Thanks! Chiefcat

[GenoMax]

There are a couple of answers about how to make these files in a prior thread. See if that is useful: Where can I get ?or how can I make a mappability track for hg38 assembly

Answer 1

[rfran010]

I personally didn't find mappability tracks for mm10 either, but there are some for mm9.

I would note that there is a major difference in blacklist regions and mappability tracks. I understand blacklist regions are for regions that can be artifacts with very high signal, while mappability focuses more on uniqueness of sequences. Also, I don't think blacklists have values associated with their regions, while mappability tracks will.

For deeptools, I think it mainly ignores the mappability value and tries to just exclude regions of low mappability (i.e. excludes regions of low signal, while blacklists exclude regions of high signal). In my experience the majority of these regions are the repetitive regions, so a better alternative than using blacklists may be using a repeatMasker track/table.