Gene Set Enrichment Analysis (GSEA)
Gene Set Enrichment Analysis · Also known as: GSEA, gene-set analysis, functional enrichment analysis, pathway-level enrichment
Gene Set Enrichment Analysis (GSEA) is a computational method that determines whether a predefined set of genes — representing a biological pathway, process, or function — shows statistically significant, coordinated differences between two biological conditions. Unlike simple fold-change filtering, GSEA operates on all measured genes ranked by a correlation metric, detecting subtle but consistent shifts across an entire pathway even when no single gene passes a significance threshold.
Read the full method
Sign in with a free account to read this section.
Method map
The neighbourhood of related methods — select a node to explore.
+27 more
When to use it
Use GSEA when you have a continuous ranked expression profile (from RNA-seq, microarray, or proteomics) comparing two clearly defined biological conditions and want to detect coordinated pathway-level changes — especially when individual genes show only modest fold-changes that would be invisible to threshold-based methods. It is particularly valuable for mechanistic interpretation of omics results and hypothesis generation. Do NOT use GSEA when you have fewer than three samples per group (permutation statistics are unreliable); when the two conditions are not cleanly separable phenotypes; or when you need absolute quantification of expression levels rather than relative directional enrichment. If the question is simply 'which pathways are overrepresented in my list of significant genes', a faster over-representation analysis (ORA / Fisher's exact test) may be more appropriate.
Strengths & limitations
- Uses all measured genes without an arbitrary significance cutoff, preserving statistical power for pathways with moderate, consistent shifts.
- The leading-edge analysis identifies the specific genes driving enrichment, providing mechanistic hypotheses beyond a p-value.
- Robust to choice of gene-set database: works with MSigDB, KEGG, Reactome, GO, and custom gene sets.
- Phenotype permutation controls for gene-gene correlations inherent in expression data, producing well-calibrated FDR estimates.
- Widely used and validated across thousands of published studies, making results directly comparable to the literature.
- Requires at least three to seven samples per group for phenotype permutation to produce a reliable null distribution; very small cohorts must fall back to the less powerful gene permutation mode.
- Results depend on the choice of ranking metric; different metrics (signal-to-noise ratio vs. log-fold-change vs. t-statistic) can produce different enrichment patterns.
- Gene sets in public databases vary greatly in curation quality; poorly curated or excessively broad gene sets inflate false-positive enrichment.
- GSEA is a between-condition method — it cannot be applied to a single sample or a continuous trait without converting the problem into a ranked-list comparison.
Frequently asked
What is the difference between GSEA and over-representation analysis (ORA)?
ORA (e.g., Fisher's exact test or hypergeometric test) requires a binary gene list — genes above a fold-change or p-value cutoff versus the rest — and asks whether pathway members are overrepresented among the 'hits'. GSEA uses a continuously ranked list of all genes and detects coordinated shifts even when no single gene is individually significant. ORA is faster and easier to interpret but is sensitive to the significance threshold; GSEA is more powerful for subtle, distributed pathway signals but requires clean phenotype contrast.
How many permutations should I use?
The standard recommendation is 1,000 permutations for exploratory analysis; 10,000 permutations are advisable when precise FDR estimates are needed for publication or when borderline gene sets are near the q < 0.25 threshold. More permutations reduce variance in the null distribution at the cost of computation time.
Can I apply GSEA to single-cell RNA-seq data?
Standard GSEA was designed for bulk expression profiles with replicated samples per condition. For scRNA-seq, pseudo-bulk aggregation per cell type followed by GSEA is the most straightforward approach. Cell-type-specific methods such as scGSEA or AUCell are alternative tools that operate at the single-cell level but have different statistical assumptions and interpretations.
Which gene-set database should I use?
The Molecular Signatures Database (MSigDB) is the most widely used and provides curated hallmark gene sets (H collection, 50 sets representing well-defined biological states), canonical pathways (C2), and gene ontology terms (C5), among others. For metabolic questions KEGG and Reactome offer mechanistic detail. Using the hallmark collection first is recommended for exploratory work because the sets are non-redundant and highly curated, reducing multiple-testing burden.
My top pathways are all immune-related — is that meaningful or a bias?
It depends on context. Immune-related gene sets are among the largest and best-curated in MSigDB, and immune processes respond rapidly to many perturbations, so genuine enrichment is common. However, very large gene sets accumulate false positives because they overlap with many other processes. Check whether the leading-edge genes are specifically relevant to your biology, and confirm enrichment with a complementary method (e.g., pathway network analysis or independent validation cohort) before drawing conclusions.
Sources
- Subramanian, A., Tamayo, P., Mootha, V. K., Mukherjee, S., Ebert, B. L., Gillette, M. A., Paulovich, A., Pomeroy, S. L., Golub, T. R., Lander, E. S., & Mesirov, J. P. (2005). Gene set enrichment analysis: A knowledge-based approach for interpreting genome-wide expression profiles. Proceedings of the National Academy of Sciences, 102(43), 15545–15550. DOI: 10.1073/pnas.0506580102 ↗
- Mootha, V. K., Lindgren, C. M., Eriksson, K. F., Subramanian, A., Sihag, S., Lehar, J., Puigserver, P., Carlsson, E., Ridderstrale, M., Laurila, E., Houstis, N., Daly, M. J., Patterson, N., Mesirov, J. P., Golub, T. R., Tamayo, P., Spiegelman, B., Lander, E. S., Hirschhorn, J. N., Altshuler, D., & Groop, L. C. (2003). PGC-1alpha-responsive genes involved in oxidative phosphorylation are coordinately downregulated in human diabetes. Nature Genetics, 34(3), 267–273. DOI: 10.1038/ng1180 ↗
How to cite this page
ScholarGate. (2026, June 3). Gene Set Enrichment Analysis. ScholarGate. https://scholargate.app/en/bioinformatics/gene-set-enrichment-analysis
Which method?
Set this method beside its closest kin and read them side by side — the library lays the books on the table; the choice is yours.
- Multi-omics gene set enrichment analysisBioinformatics↔ compare
- Network-based gene set enrichment analysisBioinformatics↔ compare
- Pathway Enrichment AnalysisBioinformatics↔ compare
- RNA-seq Differential ExpressionBioinformatics↔ compare
- Single-cell RNA-seq analysisBioinformatics↔ compare