chordoma_rna_example

RNAseq differential expression analysis pipeline using the airway dataset as a worked example.

Overview

The analysis (rnaseq_analysis.qmd) covers the following steps:

Step	Tool	Output
Differential expression	DESeq2	deseq2_results.csv
Heatmap (top 50 variable genes)	ComplexHeatmap	inline figure
Volcano plot	EnhancedVolcano	inline figure
Gene-set enrichment (GSEA)	clusterProfiler GSEA()	gsea_results.csv, inline figure
Over-representation analysis (ORA)	clusterProfiler enricher()	ora_results.csv, inline figure

Gene sets for both GSEA and ORA are the CP:KEGG_MEDICUS collection sourced from MSigDB via the msigdbr package.

Requirements

R packages

Install the required packages from Bioconductor and CRAN:

if (!requireNamespace("BiocManager", quietly = TRUE))
  install.packages("BiocManager")

BiocManager::install(c(
  "airway",
  "DESeq2",
  "ComplexHeatmap",
  "EnhancedVolcano",
  "clusterProfiler",
  "org.Hs.eg.db",
  "AnnotationDbi"
))

install.packages(c("msigdbr", "dplyr", "tibble", "circlize"))

ashr (used for LFC shrinkage) is also required:

install.packages("ashr")

Alternatively, use the provided Dockerfile to build a container, changing the information between the <> characters as needed:

docker buildx build --platform linux/amd64,linux/arm64 --push -t <dockerhub_username>/<container_name>:<container_tag> .

Or Singularity pull to use on Oscar with Open On Demand RStudio on Singularity (https://docs.ccv.brown.edu/oscar/connecting-to-oscar/open-ondemand):

singularity pull chordoma_rna_example.sif docker://jwalla12/chordoma_rna_example:quarto

Usage

Open rnaseq_analysis.qmd in RStudio (or another Quarto-aware editor) to run the analysis interactively chunk by chunk, or render it into a self-contained HTML report:

quarto::quarto_render("rnaseq_analysis.qmd")

or from a terminal:

quarto render rnaseq_analysis.qmd

All CSV output files are written to the working directory. Figures are embedded directly in the rendered HTML document.

Dataset

The airway package provides RNA-Seq read counts from an experiment on airway smooth muscle cells treated with dexamethasone (Himes et al., 2014, PLoS ONE). The contrast of interest is treated (trt) vs untreated (untrt) while accounting for cell-line (cell) as a covariate.

Analysis details

Differential expression (DESeq2)

• Design formula: ~ cell + dex
• Pre-filtering: genes with fewer than 10 total counts removed
• LFC shrinkage with the ashr method

Heatmap (ComplexHeatmap)

• Top 50 most variable genes selected after variance-stabilising transformation (VST)
• Rows Z-score normalised; columns annotated by treatment and cell line

Volcano plot (EnhancedVolcano)

• X-axis: shrunken log2 fold-change
• Y-axis: adjusted p-value
• Thresholds: |LFC| > 1, padj < 0.05

Pathway analysis (clusterProfiler + msigdbr)

• Gene sets: CP:KEGG_MEDICUS (MSigDB C2 category, Homo sapiens)
• GSEA: genes ranked by DESeq2 Wald statistic; GSEA() with eps = 0
• ORA: significant DE genes (padj < 0.05, |LFC| > 1) tested against all genes with a valid symbol as the background universe; enricher() used