Hello everybody,

I have a question about the mod-argument for ComBat (sva-package). Imagine I have two batches of different sizes with design matrices like this:

Batch 1:

ID           Group      
Sample_1     A
Sample_2     B
Sample_3     C
Sample_4     C

and Batch 2:

ID         Group
Sample_5   A
Sample_6   A
Sample_7   B
Sample_8   B
Sample_9   C
Sample_10  C
Sample_11  C
Sample_12  C

I can guarantee that batch 2 has exactly the same group-distribution as batch 1. It is however larger by a factor n. I am also certain, that some groups will only occur once in the smaller batch 1. Under these circumstances, is it strictly necessary to include the group as covariate in the design matrix for ComBat?

Many thanks in advance!