Microbiome Statistics

One-Sentence Definition

Microbiome statistics are the methods for analyzing sparse, compositional count data describing microbial community structure and its association with host or environmental variables.

Simple Explanation

Microbiome tables are mostly zeros and only give proportions — normal statistics mislead unless you account for that.

Detailed Scientific Explanation

ConceptMeaning
Alpha diversityWithin-sample richness/evenness (Shannon, Simpson, observed features)
Beta diversityBetween-sample dissimilarity (Bray–Curtis, UniFrac — phylogeny-aware)
OrdinationPCoA/NMDS visualization of beta diversity
PERMANOVATesting group differences in community composition
CompositionalityOnly relative abundances observed → CLR transforms, ALDEx2, ANCOM-BC
Differential abundanceWhich taxa differ; high method disagreement — report method explicitly
Rarefaction vs normalizationDepth control; contested

Confounders that dominate real studies: batch effects, DNA extraction kit, sequencing depth, diet, antibiotics, and low-biomass contamination (“kitome”).

Mechanism

Model counts as compositions on a simplex; use log-ratio transforms or distribution-aware models rather than raw proportions.

Clinical Importance

Research Importance

  • Reproducibility of microbiome findings hinges on these choices

Diagnostic Relevance

AMR Relevance

  • Resistome burden comparisons across populations use the same statistical framework

Active Recall Questions

  1. Alpha vs beta diversity?
  2. Why is compositional data a statistical problem?
  3. Name two major confounders in microbiome studies.

Connections