DESeq MA-plot diagnosis

Question

[Ming Tommy Tang]

Dear Biostarers,

I recently did RNA-seq analysis with DESeq with the counts from HTSeq, and the MA plot looks different from what I have before:

thousands of differentially expressed genes were identified. one thing confused me is that the red spots ( adjust pvalue <0.1) spread around. for lowly expressed genes, the log2 fold needs to be bigger to get it more significant right?

This is a simple experimental design: condition<- c("control", "control", "treatment1")

I do not have duplicate for the treatment1. I thought it was due to this, but I have duplicates for treatment2.

condition2<- c("control","control","treatment2","treatment2")

the MA plot is similar:

previously, I have another MA plot below which I think is more "right".

Any insights? Thank you so much!

Tommy

Answer 1

[Devon Ryan]

This is likely just a bug in how the colors are assigned. See here: Deseq's plotMA color-coding at random?

[Ming Tommy Tang]

Thank you Devon Ryan, I saw that post and when I reordered the res dataframe and plot with plotMA, it looked fine, but why sometimes plotMA woks fine while sometimes it works not as expected?...