ShinyMB

ShinyMB calculates sample size and statistical power for case-control microbiome studies by using simulation-based approaches to detect differences in microbial community composition and OTU-specific abundances.

Key Features:

• Statistical tests integration: Implements a recoded two-sample generalized Wald test from the 'HMP' R package for community composition comparisons and optionally uses the Wilcoxon-Mann-Whitney test for OTU-specific abundance comparisons.
• Dirichlet-Multinomial modeling: Uses the Dirichlet-Multinomial model to describe and generate microbial abundance data for simulations.
• Simulation-based power estimation: Employs simulation-based approaches to estimate statistical power across varying effect sizes and study designs.
• Sample size estimation: Estimates required sample sizes to achieve specified statistical power for detecting hypothesized differences in microbial composition.
• Stratification analysis: Evaluates the impact of stratifying samples on statistical power and on reductions in required sample size.

Scientific Applications:

• Case-control microbiome studies: Designing studies to detect differences in microbial communities between case and control groups.
• Community-level comparisons: Assessing detectability of differences in overall community composition using the generalized Wald test.
• OTU-level comparisons: Assessing detectability of differences in operational taxonomic unit (OTU) abundances using the Wilcoxon-Mann-Whitney test.
• Stratified study designs: Quantifying power gains and potential sample size reductions achieved through stratification.

Methodology:

Uses the Dirichlet-Multinomial model to simulate microbial abundances and applies simulation-based techniques to estimate power; statistical testing includes a recoded two-sample generalized Wald test from the 'HMP' R package for community composition and the Wilcoxon-Mann-Whitney test for OTU-specific comparisons.