How to control age and gender effect in expression data using EdgeR

Question

[jamieoyang]

Hi, I have multiple rna-seq samples with patients who underwent shock vs controls (Variable of interest = VAR). A sample of how the data looks like is this:

- Patient Gender Age VAR
 - 1              F      4      Shock
 - 2              M      2      Control
 - 3              M      1      Shock 
 - 4              F      6      Shock
 - 5              F      3      Shock
 - 6              F      2      Control

My current edgeR design is just:

design <- model.matrix(~VAR)

I wanted to remove/control for the effect of Gender and Age on my results. How would I write the design variable to do this? I am a beginner at this, and I have read the EdgeR manual but am still slightly confused!

I would be so grateful for any help!

-Jamie

[Nicolas Rosewick]

Before correcting for age and gender, a good idea would be to do a PCA plot and color your sample by age, gender and VAR if you see some variable that may have an impact on the sample grouping.

[jamieoyang]

That is a good idea, I will definitely do that, thank you!

After that though, do you have any advice on how to write the design model?

[Asaf]

Do you expect interactions between sex or age and VAR? Meaning do you want to subtract the effect of age or do you think age would change the response to shock?

[ATpoint]

The question is if also if it is worth bothering as n=2 for male and n=4 for female in human patients is a very small sample size given all the potential confounding effects in humans.

[jamieoyang]

Hello, thank you for your reply! The n is much larger than the 6 people I have up there, that was just provided as a sample for the variables I have for my patients.

I would like to subtract the effect of age!