What does DESeq2 non symmetric MA plot mean in differential analysis?? (hi-chip)

Question

[woongjaej]

Hi, I'm a researcher using DESeq2 for hi-chip differential analysis.

When I performed differential analysis using DESeq2, I found that my data's MA plot is far non-symmetric.(Even for not significant values)

I thought it can happen because it's not RNA-seq data. But I can't explain exactly why.

I have three questions here.

1. Does someone know what's the meaning of non-symmetric MA plot in DESeq2 for the data? Can this happen or should I think my data is not good?
2. If non-symmetric MA plot like this is not good, can I make it symmetric???
3. Can I use DESeq2 for differential loop analysis??(My data's coverage is similar between samples)

Thank you for your help.

[Ram]

A couple of points:

1. The Tool type post is for advertising/speaking about tools, not for Questions concerning tools. The category for that would be the Question category.
2. Please see How to add images to a Biostars post to add your images properly. You need the direct link to the image or the HTML embed code, not the link to the webpage that has the image embedded (which is what you have used here)

[woongjaej]

Oh, I see.

Thank you very much

[ATpoint]

Cross-posted at Bioc (it is good practice to link crossposts): https://support.bioconductor.org/p/p134337/

Answer 1

[ATpoint]

This looks like trended bias. I would guess that actually the entire plot would need to be shifted like upwards along the y-axis by +1 and by this most of these 10307 would no logner be significant. I'd say that normalization is totally off here using the default size factor method from DESeq2 which does not (afaik) handle trended biases. The farright part of the plot, kind of the "arrowhead" of this plot is probably all non-DE. Take a look at diffHic, a method developed for differential HiC analysis that might be of interest here, especially the part about removal of trended biases which is section 5.2.1 in https://bioconductor.org/packages/release/bioc/vignettes/diffHic/inst/doc/diffHicUsersGuide.pdf

[woongjaej]

Oh, thanks a lot!

I'll look into diffHic.

Thank you very much