1
$\begingroup$

Is it really required to do GC normalization ATAC seq data ?

One of the paper where they have ATAC seq data did the normaisation of GC bias after peak calling. When i looked for the library and the reason why GC bias i found this

Could it be true that genes with higher GC content are higher expressed? It has been suggested that genes that are either extremely high or extremely low expressed are under some form of selection leading to “extreme” GC content. What CQN does, is making the effect of GC content comparable across samples, and we show in [1] that this leads to improved inference. It also flattens the effect of GC content on gene expression, but we believe this is better than having the effect of GC content depend on the sample.
https://www.bioconductor.org/packages/release/bioc/vignettes/cqn/inst/doc/cqn.pdf

Now that scenario is for RNA seq would it be logical to do the same for ATAC seq?

$\endgroup$
3
  • 1
    $\begingroup$ It is not required if you ask me. GC content is the same across samples per genomic region so it would perhaps make sense if you want to compare regions within the same sample, but then other biases such as mappability would need correction as well, not sure if the bother is necessary. I personally do not do it. It is not established from what I know and I do not see the advantage. You can simply plot GC content vs peak counts to explore yourself if you like. $\endgroup$
    – ATpoint
    Dec 5 '20 at 9:34
  • $\begingroup$ " I personally do not do it. It is not established from what I know and I do not see the advantage. " i have the doubt but not the knowledge if i should do or not ! ..i came across paper where they did that but not sure why ..but now im clear how to proceed thank you and @Devon Ryan for the answer $\endgroup$
    – kcm
    Dec 7 '20 at 20:50
  • 1
    $\begingroup$ You will always find papers doing this or that. It comes down to whether the paper shows conclusively whether what they did is beneficial or necessary, and to my knowledge this has not been done for ATAC-seq towards GC normalization, whereas for RNA-seq it has been shown (see Alpine package at Bioconductor, Mike Love the developer/first author). $\endgroup$
    – ATpoint
    Dec 8 '20 at 10:14
2
$\begingroup$

The only reason to normalize for GC content in RNA-seq is if it differs notably between samples/groups. If that's not the case and you aren't trying to compare genes withing samples then you have no reason to try to account for GC content. The same goes for ATAC-seq, though there the danger with trying to normalize for GC content is that you then mask changes in global accessibility.

In short, normalize for things that should cause problems with your analysis and nothing else.

$\endgroup$
1
  • $\begingroup$ Now my confusion is cleared $\endgroup$
    – kcm
    Dec 7 '20 at 20:48

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.