Does sequencing depth influencing ATAC peak calling?

Question

[Char]

Hi All,

I was wondering if anyone had experience regarding different depths of sequencing (for ATAC sequencing) and the number of peaks you would expect to be called by HMMRATAC or MACS2/3. We have done a trial run of some atac sequencing library preparation at our lab and have sequenced those initial samples at a low level (~3.3m reads/sample). When peaks were called using HMMRATAC V1.2.10, less than 200 peaks were found per sample, which seemed worryingly low. We are trying to determine whether this is simply because we used a low sequencing depth or whether something like not having a blacklist (non-model species) or our initial library prep may have played a role?

Thank you!

[biofalconch]

Yes, basically peaks are called when reads stack above noise, when you don't have enough sequencing depth it is expected that not very abundant peaks would be available. They touch the subject in this paper, mentioning they need around 200 million reads (!!!) per sample, but people get away with less reads. It all boils down to how big your genome is, and how fast would you reach saturation if you would increase the number of reads.

[Char]

Thank you for the explanation, you've cleared up this question for me :)