Bayesian differential analysis of cell type proportions: opinion