Compositional data comprise vectors that describe the constituent parts of a whole. Data arising from various -omics platforms such as 16S and RNA-sequencing are compositional in nature. Correlations between features on raw counts have no meaningful interpretation. I previously developed metrics of proportionality for count-based compositional data. Even though these metrics can handle a modest number of zeros, they do not perform as well when the proportion of zero counts is high. Estimators for Kendall's tau for a zero-inflated models have been proposed (Pimental et al. 2015). In this talk, I will discuss the development of a version of this estimator suitable for compositional data, in particular, in the context of a modified zero-inflated logit-normal multinomial model where dependence is specified through a copula structure. I will also show results from applying this estimator on microbiome metagenomic sequencing samples from a pediatric Multiple Sclerosis study.