Pipeline for quantitative ATAC-seq

Question

[abhay.kanodia]

Hello,

I am looking for a detailed tutorial for ATAC-seq analysis using DE-Seq2 but unable to figure out the pipeline.

1. Step1: Convert MACS2 narrowpeaks file to saf. Is there another file I should use instead of MACS2 narrowpeaks? Can someone pleas tell me how to do that? I just have one untreated and one treated human cell line sample with no replicates.

2. Use featurecounts to determine the counts table. I am not sure which command (https://subread.sourceforge.net/featureCounts.html#:~:text=featureCounts%20is%20a%20highly%20efficient,and%20genomic%20DNA%2Dseq%20reads.)to use for featurecounts for ATAC-seq data. I think it also requires an annotation file which I am not sure where to get from? Is the gencode file (Comprehensive) fine to use? https://www.gencodegenes.org/human/release_42.html

Once I have this, I can use DESeq2. But I would highly appreciate your help in the initial steps.

[seidel]

A relatively simple approach: use R with GenomicRanges and rtracklayer libraries to: (1) import your narrowPeak peak files as bed files, (2) reduce() them to a common set of intervals to be quantified (the locations available to both treated and untreated), (3) read in the BAM files from each condition and count how many reads overlap the peak set from each condition. (4) now you have a count table or DE Seq. (5) optional, use distanceToNearest() and other GenomicRanges, and GenomicFeatures functions to map peaks to nearest genes, etc. (6) optional: generate coverage from your BAM files, normalize to RPM, and export to bigWig for browser views.

[abhay.kanodia]

Thanks so much @seidel! I will give it a shot!

Answer 1

[Rory Stark]

The Bioconductor packages DiffBind and csaw both have extensive vignettes/tutorials outlining a workflow for this. DiffBind uses MACS peaks and DESeq2 (or edgeR), while csaw skips the peak calling step entirely and is based on edgeR. These tutorials are useful even if you don't end up using either package as they discuss various potential issues and pitfalls.