Estimates covariation within and between two count datasets where the counts contain multinomial variation (e.g., sequence count data like microbiome 16S or bulk/single-cell RNA-seq). The model outputs Bayesian posterior samples over covariance matricies. The entire posterior reflects uncertainty in the true covariation due to multinomial counting.