MIMIX

MIMIX (MIcrobiome MIXed-effects) is a Bayesian hierarchical model for the analysis of high-throughput sequencing abundance data from designed experiments. It achieves four scientific objectives:

1. Global tests of whether experimental treatments affect microbiome composition,
2. Local tests for treatment effects on individual taxa and estimation of these effects if present,
3. Quantification of how different sources of variability contribute to microbiome heterogeneity, and
4. Characterization of latent structure in the microbiome (pictured below), which may suggest ecological subcommunities.

Grantham, NS, Reich, BJ, Borer, ET, Gross, K (2019). MIMIX: a Bayesian Mixed-Effects Model for Microbiome Data from Designed Experiments. Journal of the American Statistical Society (JASA).

Code available at github.com/nsgrantham/mimix.