Pre-ranked GSEA

Run pre-ranked gene set enrichment analysis (GSEA) in your browser. Upload a ranked gene list (e.g. DESeq2 log2 fold change or t-statistic) and a GMT gene set file, then get ES, NES, nominal p, FDR q, enrichment plots, and leading-edge genes. No account required.

WHAT IT’S FOR

You already have differential expression or another genome-wide ranking and want to know which pathways or MSigDB collections are enriched at the top or bottom of the list — without installing R, Python, or the Java GSEA desktop app.

Typical inputs: RNA-seq rank metrics, proteomics fold changes, or any signed score where higher = more up-regulated in the phenotype you care about. Gene sets come from a .gmt file you download from MSigDB (check their licence for your use) or your own curated GMT.

HOW IT WORKS
  1. Paste or upload a two-column ranked list: gene and score.
  2. Paste or upload GMT (one gene set per line: name, description, genes…).
  3. Set permutation count (default 1000), min/max gene set size, and FDR display cutoff.
  4. Click Run GSEA — enrichment scores and permutations run locally in a Web Worker.
  5. Click a result row for the running enrichment plot and leading-edge gene list.

Results use a standard pre-ranked GSEA walk (score-weighted hits vs misses) with score permutation for null distributions. Numbers may differ slightly from GSEA desktop or GSEA_R; treat this as a fast exploratory tool and validate key hits independently.

DATA & PRIVACY

Ranked lists and GMT files are processed entirely in your browser. Nothing is uploaded to our server. Do not paste identifiable patient data unless your policy allows. See the privacy policy for site-wide analytics and cookies.

Two columns: gene symbol (or ID) and score. Higher score = stronger up-regulation. TSV or CSV. Header row optional (gene, score).

Standard GMT format: SET_NAME <tab> description <tab> GENE1 <tab> GENE2 … Download collections from MSigDB.

Demo uses min size 5. For MSigDB Hallmark, use min 15 / max 500 to match common GSEA defaults.

About the author: Dr Mark Bond, The Bond Lab, University of Bristol. Contact · mark.bond@bristol.ac.uk ORCID.