Should BAM files be Tn5-shifted before using featureCounts after Genrich peak calling?

[wangbenwang123]

I used Genrich to call peaks separately for two experimental conditions (e.g., control vs. treated) and then merged the peaks into a unified peak set (merged_peaks.bed) for downstream quantification using featureCounts.

Since ATAC-seq requires Tn5 transposase shift correction (where + strand reads should be shifted +4 bp and - strand reads should be shifted -5 bp), I am unsure whether featureCounts should be run directly on the original BAM files or whether I need to adjust the BAM files beforehand.

My Questions
Should I use the original Genrich BAM files for featureCounts, or do I need to shift the BAM alignments before quantification?

[jsh58]

It is unclear to me why you are using featureCounts here. If you want the counts of reads in peaks, Genrich supplies that information in its output(s). Genrich already corrects for Tn5 transposase occupancy, as described here.