Nucleoatac - extending broadPeak regions

Question

[BioinfGuru]

I am about to use nucleoatac and read the following by the authors regarding the broad peaks called by MACS2: "It is potentially advisable to extend these regions a bit (e.g. using bedtools slop)"

From a 2016 paper using nucleoatac: "peak regions were widened to 2500 bp, centered over the maximum read density value calculated, and used as input regions of interest for NucleoATAC analysis"

Does anyone have nucleoatac experience with this? Nucleoatac implements a mixture model based on the overall fragment length distribution of a given sample. The occupancy measures how likely a given fragment (0-1 scale) is to come from nucleosome-free (NFR) DNA or nucleosomal DNA.

MACS2 output:

fragment length = 200
min broad peak length = 200
max broad peak length = 5000

I'm guessing that the idea is to extend the short peaks, but this is just a guess and I cant defend why.

Thank you,

Kenneth

Answer 1

[igor]

According to the developer (regarding "extend these regions a bit"):

By a bit I meant 100-200 bp. Main downside with extending by a lot is increased runtime-- in regions with fewer reads, there is less power to detect nucleosomes so it is likely a waste of time. It will also dilute quality of nucleosome positioning calls. Already when looking within peaks you want to keep in mind that absence of nucleosome call does not mean absence of nucleosome (could just be insufficient reads).

Source: https://github.com/GreenleafLab/NucleoATAC/issues/46

[BioinfGuru]

well spotted. thank you :)