How to perform DESEQ2 analysis when there are three treatments (one of them being a reference)?

Question

[mrj]

I am trying to analyze a gene expression dataset with three treatments (A, B, and control). There should not be any interaction effect. I will be thankful to you if can briefly explain how to obtain differentially expressed genes under these conditions.

How can I define my reference for the expression analysis when I have three treatments? How can I define my formula?

is it ~ A + B?

when I show over-expressed genes, should I have two lists? e.g.,

A vs Control B vs Control

Answer 1

[Kevin Blighe]

Hi, you should have a single column in your metadata for condition, and this should contain A, B, and control.

For example:

metadata <- data.frame(
  condition = c(
    'A','A','A',
    'B','B','B',
    'control','control','control'))
metadata$condition <- factor(metadata$condition,
  levels = c('control','A','B'))

So, your design formula would be:

~ condition

Then, you can compare different levels like this: DESeq2 compare all levels

Kevin

[mrj]

Thank you so much Kevin. This helps a lot.