The rise to power of the microbiome: power and sample size calculation for microbiome studies

Abstract

A priori power and sample size calculations are crucial to appropriately test null hypotheses and obtain valid conclusions from all clinical studies. Statistical tests to evaluate hypotheses in microbiome studies need to consider intrinsic features of microbiome datasets that do not apply to classic sample size calculation. In this review, we summarize statistical approaches to calculate sample sizes for typical microbiome study scenarios, including those that hypothesize microbiome features to be the outcome, the exposure or the mediator, and provide relevant R scripts to conduct some of these calculations. This review is intended to be a resource to facilitate the conduct of sample size calculations that are based on testable hypotheses across several dimensions of the microbiome. Implementation of these methods will improve the quality of human or animal microbiome studies, enabling reliable conclusions that will generalize beyond the study sample.