DEBrowser

Interactive differential expression analysis & visualization for count data.

DEBrowser turns RNA-Seq differential expression analysis into an interactive, point-and-click workflow. It wraps three established Bioconductor engines — DESeq2, edgeR, and limma — in a Shiny app, so changing a cutoff, a normalization method, or a comparison re-draws every plot and table in real time. No code required to explore your results; full reproducibility exports when you're done.

📖 Full user guide: the complete, screenshot-by-screenshot walkthrough lives at debrowser.readthedocs.io. It is hosted outside the package to keep the package itself small.

Highlights

• Guided six-step wizard — Quick start → Upload → Filter & normalize → Batch effect → Comparison → DE analysis. Each step unlocks the next.
• Three DE engines — DESeq2, edgeR, limma, with their key parameters exposed in the UI.
• Interactive Main Plots — scatter / volcano / MA with Up/Down/NS coloring; lasso- or box-select genes to spawn a linked heatmap.
• Quality control — all-to-all correlation, PCA, IQR, and density views, before and after batch correction.
• Cross-contrast concordance — see where two or more comparisons agree.
• Enrichment — GO/KEGG over-representation (clusterProfiler) and GSEA (fgsea) with a leading-edge and NES-heatmap view.
• Reproducibility — export the session as an R script, R Markdown / HTML, Jupyter notebook, or a ready-to-paste methods paragraph.
• Bookmark & share the full analysis state behind a stable URL.
• Optional AI interpretation of gene-set biology (off by default; local Ollama supported for privacy).
• Modern, restrained UI — a "Clinical Indigo" theme with light/dark modes and 1–6 / T keyboard shortcuts.

Installation

# From Bioconductor (release)
if (!requireNamespace("BiocManager", quietly = TRUE))
    install.packages("BiocManager")
BiocManager::install("debrowser")

library(debrowser)
startDEBrowser()

startDEBrowser() opens the app in your browser on port 3838 (so bookmark URLs stay stable across restarts). See Operating system dependencies if system libraries are missing.

New to DEBrowser? Click Vernia et al. under Demos on the Upload step, then Upload, to walk the whole pipeline on bundled data.

Walkthrough

1. Load your data

Drop in a count matrix (genes in the first column, one raw-count column per sample) and, optionally, a metadata table mapping samples to conditions and batches. DEBrowser accepts .csv / .tsv / .txt / .csv.gz. On upload it shows a summary — samples, genes, conditions — plus a preview and the sample-design table.

DESeq2 needs un-normalized counts (it models library size internally). Use pre-normalized values only with edgeR or limma.

2. Filter & normalize

Trim low-signal features with Max, Mean, or CPM filtering. Row counts before and after are shown side by side, with per-sample histograms.

3. Correct batch effects (optional)

Choose a normalization (MRN, TMM, RLE, upperquartile) and a correction method (ComBat, ComBat-Seq, Harman). Inline PCA / IQR / Density plots — Before vs. After — confirm samples cluster by biology, not batch.

4. Choose comparisons

Define one or more contrasts by assigning samples to each side. Each becomes its own DE result set you can switch between.

5. Run DE and explore

Start DE runs the analysis (progress is reported in stages) and opens Main Plots. Genes are colored Up (red), Down (blue), NS (grey) per your padj and log2-fold-change cutoffs; every change is instant.

Switch to the Volcano or MA view with the same controls; lasso- or box-select a region to spawn a linked heatmap of just those genes.

Beyond DE

• QC Plots — all-to-all correlation, PCA, IQR, density across the dataset.
• Concordance — where two or more comparisons agree/disagree.
• Enrichment — GO/KEGG (enrichGO, enrichKEGG, Disease, compareCluster) and GSEA against MSigDB, with a NES heatmap.
• Tables — searchable, sortable result tables (All Detected, Up/Down, Most varied, Comparison differences) with regex search that applies everywhere.

Reproducibility & sharing

The Export menu turns your interactive session into an R script, R Markdown / HTML, a Jupyter notebook, or a methods paragraph for your manuscript. The Bookmark button captures the entire analysis state behind a stable, shareable URL.

AI interpretation (optional)

DEBrowser can summarize the biology of a GSEA gene set on the Enrichment tab. It is off by default — no network calls until you enable it and configure a provider (Anthropic, OpenAI, or local Ollama). Per-call privacy modes control what leaves your machine; API keys are stored in your OS keychain via keyring, never in plaintext.

Install the AI extras once: install.packages(c("ellmer", "whisker", "keyring")).

Design system

DEBrowser ships a self-contained visual redesign layer. Contributors editing the UI should read DESIGN.md (the token vocabulary + editing rules) and open the live component gallery with debrowser::de_style_guide(). Full rationale: docs/design/.

Operating system dependencies

• Fedora / Red Hat / CentOS: openssl-devel libxml2-devel libcurl-devel libpng-devel
• Ubuntu: sudo apt-get install libcurl4-openssl-dev libssl-dev libxml2-dev libudunits2-dev

Documentation

• Full user guide (illustrated, with worked examples): debrowser.readthedocs.io — source in the debrowser-docs repo.
• Quick reference: vignette("DEBrowser"), or the Bioconductor page.
• Design system & contributor notes: DESIGN.md, docs/design/.

Citation

If you use DEBrowser in your research, please cite:

Kucukural A, Yukselen O, Ozata DM, Moore MJ, Garber M. DEBrowser: interactive differential expression analysis and visualization tool for count data. BMC Genomics 2019, 20:6. doi: 10.1186/s12864-018-5362-x

License

GPL-3 (see LICENSE).