I have the following experimental design (n=9). Trying to compare the two cell types.

Celltype 1 from individuals A, B, and C

Celltype 2 from individuals A, B, D, E, F, and G

Is there any way to include in the design the fact that 2/3 cell type 1 samples have a corresponding paired sample in cell type 2?

Right now I have just been running design=~Celltype. But if the design could include Individual in some way, I think that would improve the power a lot. 

Willing to remove the sample from individual C if needed.

I tried running without and without individual C, using design=~Celltype + Celltype:Individual and removing all-zero columns from the model matrix. But this produces the following error:

Error in checkForExperimentalReplicates(object, modelMatrix) :

same number of samples and coefficients to fit with supplied model matrix

So maybe this just doesn't make sense to do?

The short answer is that we can’t do this with fixed effects but you can do it with limma and duplicateCorrelation(). There are some longer posts on here with this question but it’s difficult to dig them up based on limited search functionality.