featureCounts allowMultiOverlap - fractional or not?

Question

[Lucy]

Hi,

I am running featureCounts on my ATAC-seq data. I have allowed multi-overlapping reads (i.e. reads that overlap with more than one peak). However, I am unsure whether to include the --fraction option (assign fractional counts to features) or whether I should allow each overlap to count as 1.

Does anybody have thoughts on this? What is most commonly used in this situation and what are the pros/cons of each approach?

Many thanks,

Lucy

[ATpoint]

Remember that ATAC-seq relies on the integration of a transposon into open chromatin, introducing an adapter that flanks the genomic fragments. Therefore, it is actually only the ends of the paired-end fragments that are of interest to infer chromatin accessability, and therefore, every overlap should count. Still, I would be surprised if this makes a notable difference.

[Lucy]

Thank you - I am now trying to perform featureCounts with the cut sites instead.