Differential Abundance & Compositional Statistics — Hands-on

Move beyond naive relative abundances and p values to compositional aware, publication grade differential abundance analysis. This module covers normalization, log ratio transforms, popular tools such as DESeq2, edgeR, ALDEx2 and ANCOM-BC, and how to report results with clear effect sizes and appropriate covariate handling for microbiome studies.

Differential Abundance & Compositional Statistics

Help Desk · WhatsApp

Session Index

Session 1 — Compositional Data & Normalization Session 2 — Classical DA: DESeq2/edgeR Style Models Session 3 — Compositional DA: ALDEx2, ANCOM-BC & Log Ratios Session 4 — Mini Capstone: From Counts to Interpretable DA Results

Session 1

Fee: Rs 8800

Compositional Data & Normalization

Why microbiome data are compositional

counts, proportions and closures relative vs absolute abundance subcompositional incoherence and pitfalls

Normalization options for count tables

library size factors TSS, RLE, TMM ideas rarefaction vs model based normalizations

Intro to log ratio thinking

CLR, ALR and ILR concepts pseudocounts and zeros base and reference choices

Session 2

Fee: Rs 11800

Classical DA: DESeq2/edgeR Style Models

Count models for feature wise testing

negative binomial frameworks dispersion estimation shrinkage and contrasts

DESeq2 and edgeR workflows for microbiome data

design formulas and covariates model diagnostics and MA plots p values, FDR and log2 fold changes

Limitations of RNA style DA when data are compositional

sensitivity to global shifts false gain and loss effects when these methods are still useful

Session 3

Fee: Rs 14800

Compositional DA: ALDEx2, ANCOM-BC & Log Ratios

ALDEx2 style effect estimation

Monte Carlo Dirichlet sampling median effect and uncertainty interpretation of effect statistics

ANCOM and ANCOM-BC style approaches

log ratio based testing idea bias correction and structural zeros suitable use cases and caveats

Custom log ratio based features and balances

pairwise and group log ratios balances and phylogenetic balances overview using ratios in simple regression models

Session 4

Fee: Rs 18800

Mini Capstone: From Counts to Interpretable DA Results

Designing a DA analysis for one microbiome question

guided theory plus practical

Running at least two DA approaches on the same dataset

DESeq2 or edgeR as baseline one compositional method (ALDEx2 or ANCOM-BC) comparing overlap and differences

Deliverables: volcano plots, effect size tables and narrative

ranked features with log fold change and FDR annotated figures for presentations short report explaining methods and choices

PDF

Differential Abundance & Compositional Statistics Training | DESeq2, ALDEx2 & ANCOM-BC

Differential Abundance & Compositional Statistics — Hands-on