For ATAC-seq bias correction, is an naked DNA sequencing data necessary?

[Honchkrow]

Hello,
Recently I use pyDNase to deal with my ATAC-seq data. I read the paper about Wellington-bootstrap, it is said that the bias is corrected using observed cuts and expected counts. I also read the paper about how to calculate the "predicted count". The "predicted count" can be from chromatin derived DNase-seq or naked DNA. Which method pyDNase support? Is an naked DNA sequencing data necessary?
Thanks!

[jpiper]

Hi there, naked DNA sequencing isn't necessary. Data from He. et al (2014) (Shirley Liu's lab) Naked DNA was used to determine DNase I cutting preference for all 6-mers, and this data is here

However, for ATAC-seq data if you wanted to normalise against ATAC-seq insertion bias then you'd need a dataset to determine this from. A common method is to use all the cuts outside the peaks but the better method is to use Naked DNA for this.

You can look at dnase_bias_estimator.py on how to calculate your own bias file, and then you use this fie in dnase_average_profile.py
or dnase_to_javatreeview.pywith the "--bias-file" command.

If you look in the code, what is actually happening here is we're applying the formula developed in the original bias paper.

I never got around to correcting for bias during footprinting itself before I finished my PhD. I could look into doing this, though.

[Honchkrow]

Thanks for your reply!