RNA-seq Pipeline Walkthrough

Step through a full DESeq2 differential expression analysis interactively, from count matrix to volcano and MA plots

Purpose

RNA-seq analysis has enough moving parts – normalisation, dispersion estimation, shrinkage, multiple testing, filtering – that a static tutorial rarely communicates how each step changes the final gene list. The RNA-seq Pipeline Walkthrough guides a reader through a complete DESeq2 analysis, one stage at a time, with every intermediate table and plot inspectable.

User inputs

• Count matrix (user-uploaded or a built-in GEO example)
• Sample metadata (sample ID, condition, optional covariates such as batch)
• Design formula, constructed through dropdowns
• Contrast selection: which condition vs. which reference
• Filtering thresholds: minimum count, minimum samples
• LFC shrinkage method: apeglm, ashr, normal

Outputs

• Sample-sample distance heatmap and PCA plot for QC
• Dispersion plot with fitted trend
• Results table with shrunken log-fold changes, standard errors, and adjusted p-values
• Volcano plot and MA plot with user-adjustable significance thresholds
• Gene-level count plot for the top hits
• The full DESeq2 script reproduced in a copy-ready code panel

Didactic value

Changing the LFC shrinkage method in a dropdown and seeing the volcano plot reshape – noisy low-count genes pulled toward zero, high-confidence genes barely moved – teaches what shrinkage is for in a way the DESeq2 manual never quite manages. Adding a batch covariate and watching surprising hits collapse communicates the value of design formulas.

Embedded in

• bioinformatics/bulk-rnaseq-differential-expression.md
• bioinformatics/rnaseq-quality-control.md
• bioinformatics/lfc-shrinkage.md

Source code

Local: apps/12-rnaseq-pipeline/

Run with:

shiny::runApp("apps/12-rnaseq-pipeline")