Compositional Data Analysis & Statistical Modeling — Hands-on

Learn how to correctly analyse microbiome count and relative-abundance data using compositional data analysis principles. This module covers log-ratio transforms, Aitchison geometry, differential abundance methods, multivariate modeling and robust visualisation for amplicon and shotgun feature tables.

Compositional Data Analysis & Statistical Modeling

Help Desk · WhatsApp

Session Index

Session 1 — Compositional Thinking & Aitchison Geometry Session 2 — Log-Ratio Transforms & Distance Measures Session 3 — Differential Abundance & Multivariate Models Session 4 — Mini Capstone: Compositional Analysis Report

Session 1

Fee: Rs 8800

Compositional Thinking & Aitchison Geometry

Why microbiome data are compositional

sequencing depth and closure constraint spurious correlations in relative abundances when proportions mislead inference

Aitchison geometry basics for compositions

simplex vs Euclidean space perturbation and powering operations subcomposition and amalgamation effects

Exploring raw counts vs relative abundance tables

library size distributions zero inflation and sparsity patterns quick EDA and sanity checks

Session 2

Fee: Rs 11800

Log-Ratio Transforms & Distance Measures

Handling zeros and count normalisation

pseudocount strategies and caveats filtering low prevalence features from counts to proportions safely

Log-ratio transforms for microbiome features

CLR, ALR and ILR transforms choice of reference and balances visualising log-ratio transformed data

Aitchison distances and ordination under CoDA

Aitchison distance vs Bray-Curtis PCA/PCoA on CLR or ILR space interpreting loadings and balances

Session 3

Fee: Rs 14800

Differential Abundance & Multivariate Models

Differential abundance with compositional methods

ALDEx2 style approaches ANCOM and robust log-ratio testing contrast with DESeq2 and edgeR assumptions

Regression and classification using log-ratios

linear and logistic models in CLR space sparse regression and feature selection interpreting coefficients as log-ratio effects

PERMANOVA and constrained ordination with CoDA-aware inputs

PERMANOVA on Aitchison distances CCA/RDA with transformed features linking community variation to covariates

Session 4

Fee: Rs 18800

Mini Capstone: Compositional Analysis Report

End to end CoDA workflow on a microbiome dataset

theory plus guided practical

Key plots and diagnostics for stakeholders and reviewers

Aitchison-based ordination plots volcano/MA plots for log-ratio DA effect sizes with confidence intervals

Deliverables: transformed feature tables, scripts and narrative report

CLR/ILR transformed matrices R or Python analysis notebook PDF or HTML CoDA-based microbiome report

PDF

Compositional Data Analysis & Statistical Modeling Training | CoDA for Microbiome Data

Compositional Data Analysis & Statistical Modeling — Hands-on