A contrast with four simultaneous conditions for DESeq2?
1
0
Entering edit mode
@jansoderman-11752
Last seen 3.9 years ago

Hi,

To identify genes that respond differently to inflammation in females and males (with ulcerative colitis) I have set up the following contrast using makeContrasts of the limma package, and analyzed it using limma and edgeR: (UlcerativeColitis.Inflamed.Male-UlcerativeColitis.notInflamed.Male)-(UlcerativeColitis.Inflamed.Female-UlcerativeColitis.notInflamed.Female).

How can this be accomplished for DESeq2? As I understand it the ‘contrast’ argument of the results function of DESeq2 can be provided a list, but the list should have a length of two. I would need a list of length four for this contrast. Can someone please help me solve this?

Sincerely,

Jan

DESeq2 contrast results • 1.0k views
ADD COMMENT
0
Entering edit mode
@mikelove
Last seen 3 hours ago
United States

Do you have a reason for switching among three packages to analyze the dataset? I'd recommend to stick with one statistical package, unless you have a methodological reason for changing. Trying out multiple methods on the same dataset can be problematic in terms of type I error control.

In order to contrast terms in DESeq2 you use contrast=list(c("x","y","z"), c("a","b","c")) which will put x,y, and z in the numerator and a,b, and c in the denominator.

ADD COMMENT
0
Entering edit mode

Hi Michael,

Thanks for your reply. I have a follow-up question.
First, however, I will comment on the reasons for switching among three packages. New to RNA-seq analysis and without a strong preference based on literature I started out with limma. However, I also wanted to explore the workflow of other packages and to assess to what extent my results could be confirmed using a package based on a different framework compared to limma's data transformation. I was able to easily fit edgeR into what I had already done. Later I read a paper describing a comparatively large number of genes that responded differently to inflammation in females compared to males with ulcerative colitis (using DESeq2), whereas I had not found any such genes (using limma or edgeR). I wanted to check if this discrepancy is due to the data itself (i.e. patient samples) or due to the use of DESeq2 (vs. limma or edgeR) for data analysis.

If I have the correct set-up for the contrast I still get the same result, i.e. no genes that respond differently to inflammation in females compared to males.

Since I investigate a difference of differences (i.e. (a/x)/(b/y)) I thought I would provide the following arguments:
contrast = list(c("a", "x"), c("b", "y"))
listValues = list(c(1, -1), c(1, -1)) or listValues = list(1, -1, 1, -1)
which, however, is not allowed.

Because of this I mathematically rearranged the expression (a/x)/(b/y) to its equivalence (a*y)/(b*x), and provided the argument:
contrast = list(c("a", "y"), c("b", "x"))
Is this the way to handle a difference of differences using DESeq2?

Sincerely/ Jan

ADD REPLY
0
Entering edit mode

Yes, you got it.

ADD REPLY

Login before adding your answer.

Traffic: 796 users visited in the last hour
Help About
FAQ
Access RSS
API
Stats

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.

Powered by the version 2.3.6