Microbiome Statistics: Compositional Data & Bias Correction — Hands-on

Build statistically rigorous, bias-aware microbiome analyses from the ground up. This module focuses on compositional data principles (Aitchison geometry) , correct normalizations and transformations, robust differential abundance, and thorough handling of batch effects, zeros, confounders, and technical bias. You will implement analyses in R/Python/QIIME 2 and deliver a reproducible mini-report with checklists and interpretation notes.

Microbiome Statistics: Compositional Data & Bias Correction

Help Desk · WhatsApp

Session Index

Session 1 — Compositional Foundations & Transforms Session 2 — Distances, Ordinations & Hypothesis Tests Session 3 — Differential Abundance & Confounding Session 4 — Batch/Bias Correction & Reporting

Session 1

Fee: Rs 8800

Compositional Foundations & Transforms

Aitchison geometry & pitfalls of rarefaction/TSS-only views

closure & subcompositional coherence library-size confounding feature prevalence filtering

Transforms & balances

CLR/ALR/ILR SBP/phylogenetic ILR (PhILR) balance trees & interpretation

Zeros & limits of detection

multiplicative/Bayesian replacement pseudo-count strategies sensitivity analyses

Session 2

Fee: Rs 11800

Distances, Ordinations & Hypothesis Tests

Distance choices & ordination

Aitchison (Euclidean on CLR) robust PCA (DEICODE/RPCA) UniFrac vs composition-aware

Group comparisons & covariates

PERMANOVA/permdisp caveats AN (C) OVA on balances effect sizes & CIs

Power/size for compositional analyses

simulation & Dirichlet-multinomial sparsity & prevalence impacts multiple testing control

Session 3

Fee: Rs 14800

Differential Abundance & Confounding

DA methods (strengths & caveats)

ANCOM-BC / ALDEx2 corncob / Songbird (QLR) W-statistics & rankings

Covariates & mixed models on balances

MaAsLin2 with CLR/ILR random effects (subject/time) stratified/blocked analyses

Sensitivity & robustness checks

zero-handling variants rare taxa vs aggregates simulation-based validation

Session 4

Fee: Rs 18800

Batch/Bias Correction & Reproducible Reporting

Batch & unwanted variation

ComBat/ComBat-Seq; RUV-III-NB removeBatchEffect (limma) percentile normalization

Technical bias awareness

PCR/primer & lysis biases spike-ins & mock communities contamination checks

Reporting, checklists & exports

Quarto notebooks & model cards tables/figures + env files assumptions & limitations

PDF

Microbiome Statistics: Compositional Data & Bias Correction Training | CLR/ILR, DEICODE, ANCOM-BC/ALDEx2, ComBat/RUV

Microbiome Statistics: Compositional Data & Bias Correction — Hands-on