Differential communication with a more complex design

[Hrovatin]

I would want to find differential pathways between cell type pairs for a design of ~condition+batch. Is there a way to do this? E.g. can I directly compare probabilities computed on each datasets for each pathway-cell_type1-cell_type2 combination, e.g. using linear regression with the above design? Or are the pathway probabilities not directly comparable across samples?