I am trying to perform a differencial expression analysis between two groups of patients. But I would like to group them taking into account two variables.
Imagine I have these data:
patient1 T1 6 patient2 T1 2 patient3 T4 2 patient4 T2 0 patient5 T3 1 patient6 T2 4 patient7 T3 5 patient8 T1 0
Would it be possible to group patients in two groups (for example:
late) taking into account both variables? As an example: early patients could be those presenting
T2 status referring to the first variable and a value shorter than 4 referring to the second variable. I wouldn't like to be subjective, so I would like to use some program which could perform this grouping.
On the other hand, would it be possible to create a design matrix with the two variables and then use
limma to perform the differencial expression analysis taking into account both variables to group patients?
Are you asking about software limitations, or the statistical soundness of doing what you propose?
About a method to do this with statistical soundness
I would suggest you to use
Rsubreadsto create your matrix just inputting the corresponding samples for each group and then use
EdgeRfor DE. But I think that you could use
RSEMto generate your count matrix for using it with
limma, that should be fine!