Limma/edgeR Design matrix for gender effect

Question

[kailsen]

Hi,

I'm trying to do an RNA-seq analysis using edgeR and want to negate the contribution of gender in my samples. Control (3 Male, 1 Female), Treated (4 Male, 1 Female). I want to find DE genes as a result of treatment. I wanted to know if my design file and contrast.matrix file makes sense?

 group$group
1  WT_F
2  WT_F
3  WT_M
4  WT_F
5  RX_F
6  RX_F
7  RX_M
8  RX_F
9  RX_F

design file:

  groupRX_F groupRX_M groupWT_F groupWT_M
1         0         0         1         0
2         0         0         1         0
3         0         0         0         1
4         0         0         1         0
5         1         0         0         0
6         1         0         0         0
7         0         1         0         0
8         1         0         0         0
9         1         0         0         0

attr(,"assign")
[1] 1 1 1 1
attr(,"contrasts")
attr(,"contrasts")$group
[1] "contr.treatment"

contrast.matrix:

           Contrasts
Levels      RXvsWT
  groupRX_F      1
  groupRX_M      1
  groupWT_F     -1
  groupWT_M     -1

[ATpoint]

Did you check by PCA or MDS if there is some evidence for this batch effect (I guess not simply because variation between human individuals is probably way more confounded by factors such as life style, health status, diet, consumption of drugs (medical, alcohol, cigarets) etc. = did the females cluster away from the bulk of the males?

[kailsen]

Yes, PCA showed that the males samples were away from the rest of the group.

Answer 1

[swbarnes2]

With only 9 samples, and only 1 male in each group, I don't think it makes sense to think that you can tease out the effect of sex on your samples.

[kailsen]

I m not interested in the effect of Sex on the samples but correct for the odd Male samples in my group. For now treating like a batch effect and I'm not sure if it's right.

[swbarnes2]

You can't tell whether that one sample is different because its male, or is different because it's a different individual.