henhaobio
Sign in

The InnovationBioinformaticsPeer-reviewed journal article

clusterProfiler: An universal enrichment tool for functional and comparative study

Tianzhi Wu, Erqiang HuiD, Shuangbin XuiD, Meijun CheniD, Pingfan Guo, Zehan Dai, Tingze Feng, Lang ZhouiD, Wenli TangiD, Li ZhaniD, Xiaocong Fu, Shanshan LiuiD, Xiaochen BoiD, Guangchuang YuiD

PDFDOIPreprintCodeData
In-depth full-text digest available · Jump to digest ↓

Abstract

I present the open-source software package clusterProfiler (https://github.com/GuangchuangYu/clusterProfiler) for functional enrichment analysis. clusterProfiler provides a universal interface for functional enrichment analysis for internal supported ontologies/pathways as well as annotation data provided by users or obtained from online databases. It enables comparative analysis and offers comprehensive visualization tools for result interpretation.

In-depth summary

Claude Opus 5.5 · max effort · 8 Nov 2026 · from the full text of version 2 (1 Feb 2018)

clusterProfiler performs over-representation (Fisher's exact test) and gene set enrichment analyses on Gene Ontology, KEGG, and user-supplied annotations, supporting non-model organisms, multi-condition comparisons, and ggplot2-based visualization.

Most enrichment tools target model organisms or specific domains with built-in GO or KEGG annotations; a prior study found about 42% of tools were over five years out of date, capturing only 26% of the processes or pathways that current annotation recovered. clusterProfiler is designed around up-to-date annotation (Bioconductor OrgDb, the KEGG API, UniProt) plus arbitrary user gene sets, so non-model species and other ontologies can be tested and compared across conditions.

Key results

  1. ClaimedProvides one enrichment interface with built-in GO and KEGG support plus modular sub-packages (DOSE, ReactomePA, meshes) for disease, Reactome and Medical Subject Headings (MeSH) analyses, and uses GOSemSim semantic similarity to collapse redundant enriched GO terms.

    KEGG Pathway and Module data come directly from the KEGG API, covering more than five thousand genomes plus KEGG Orthology for metagenomic work; GO annotation comes from Bioconductor OrgDb packages or from AnnotationHub, biomaRt or UniProt; GOSemSim offers information-content and graph-based measures. No benchmark against other tools. · Abstract; Abstract

  2. ClaimedGeneric enricher (Fisher's exact test) and GSEA (gene set enrichment analysis) functions accept user-defined gene-to-term mappings, extending the same workflow to newly sequenced species, ontologies lacking built-in support, and custom or downloaded gene sets.

    Named inputs include electronic annotation from Blast2GO or KAAS, InterPro domains, Clusters of Orthologous Groups, mouse phenotype ontology, GMT files of Molecular Signatures Database gene sets, and whole-genome GO pulled from UniProt by taxonomic ID. No worked example or accuracy data given. · Abstract; Abstract

  3. ShownThe ggplot2-based enrichplot sub-package goes beyond top-term bar charts, drawing gene-concept networks (standard and circular), gene-concept heatmaps with expression overlay, GO directed acyclic graphs, enrichment maps, dot plots, and GSEA ridgeline and running-score plots.

    Fig. 1a to 1i: enrichment-map edges are weighted by gene-set overlap ratio so overlapping sets group into functional modules; ridgeline plots default to core enriched (leading-edge) genes; bar charts colour terms by scores such as p.adjust, dot charts add a size-encoded score, and both can facet sub-ontologies. Illustrative outputs only; no quantitative evaluation. · Abstract; Abstract; Figure legends

  4. SuggestedcompareCluster contrasts functional profiles across experiments, conditions or time points via R-style formula syntax, at gene or genomic-region level, with region-based (BED) enrichment handled by combining clusterProfiler with the in-house ChIPseeker package.

    Fig. 1j compares disease associations of multiple genome-wide region sets from samples GSM1295076, GSM1295077, GSM749665 and GSM749666; neither the formula interface nor a time-course design is demonstrated, and no comparison statistics are reported. · Abstract; Abstract; Figure legends

  5. ClaimedReanalysis of the GTEx dataset with clusterProfiler reportedly uncovered many new pathways and hypotheses, offered as an example of how outdated annotation in earlier analyses can hide functional signal.

    Only a link to a GitHub repository (enrichment4GTEx_clusterProfiler) supports this; no pathway counts or statistics appear in the text. Cited context: a previous study found about 42% of tools outdated by more than five years, with only 26% of processes or pathways captured relative to up-to-date annotation. · Abstract

Methods, data and reproducibility

clusterProfiler applies Fisher's exact testing and gene set enrichment analysis to gene lists, or to genomic regions via ChIPseeker, against interchangeable annotation sources (OrgDb, KEGG API, UniProt, GMT files, custom sets), adds compareCluster for multi-condition comparison, and plots via enrichplot/ggplot2; no benchmarks are reported.

  • Genome-wide regions of interest (ROIs) from the listed samples, used in Fig. 1j to illustrate comparing disease associations across experiments · GSM1295076; GSM1295077; GSM749665; GSM749666
  • GTEx dataset (attributed in the text to the ENCODE consortium) reanalyzed with clusterProfiler; results available only via a linked GitHub repository
  • KEGG Pathway, Module and KEGG Orthology annotations retrieved through the KEGG API as enrichment reference data · multiple species · n = more than five thousand genomes
Methods & tools
clusterProfiler (enricher for Fisher's exact test, GSEA, compareCluster), enrichplot (ggplot2-based visualization sub-package), DOSE (Disease Ontology, Network of Cancer Gene, gene/variant-disease associations), ReactomePA (Reactome pathway enrichment), meshes (Medical Subject Headings enrichment), ChIPseeker (region annotation enabling BED-based epigenomic enrichment), GOSemSim (GO semantic similarity for redundancy removal), ggplot2 (grammar-of-graphics plotting), KEGG API (Pathway, Module and KEGG Orthology retrieval), Gene Ontology via Bioconductor OrgDb, AnnotationHub, biomaRt or UniProt, Molecular Signatures Database gene sets (GMT import), Blast2GO and KAAS (electronic annotation for newly sequenced species)
Reproducibility gaps
No software versions, named code licence, installation steps or runnable example appear in the text; the GTEx reanalysis exists only as a linked repository, and example inputs for Fig. 1 are identified only for panel j.
Limitations (4)
  • The authors raise annotation currency as a field-wide caveat rather than a limit of their tool: stale annotation severely underestimates functional significance, and its effects can persist for years and hinder follow-up studies. · stated by the authors
  • As a field-wide caveat motivating enrichplot and GOSemSim, the authors note that top-term bar charts can be dominated by redundant terms, fail to reveal complex gene-pathway links, and leave results open to annotation or interpretation bias. · stated by the authors
  • No benchmarking or quantitative validation is reported: features are not compared with other enrichment tools, the GTEx reanalysis gives no pathway counts, and GOSemSim redundancy reduction is described only for GO, not KEGG, Reactome, MeSH or user gene sets. · noted in this summary
  • Core statistical settings go unstated, including the background gene universe for Fisher's exact tests, the multiple-testing correction behind the p.adjust values shown in plots, and GSEA ranking or permutation choices, limiting assessment of result calibration. · noted in this summary

Read this if you want a short overview of clusterProfiler's supported annotations, custom gene-set inputs, multi-condition comparison and plot types before running enrichment on gene lists or genomic regions.

Citations
12,469
FWCI
592.38
Field percentile
top 1%

Reproducibility ledger

Nobody has reported trying to reproduce this paper yet. Did you replicate an experiment or run the code? Your report saves the next lab a week.

Sign in to file a reproducibility report.

Related papers

Similarity Map
SelectedSame fieldCross-field
THISGu '22Brinkmann '26Koziara '16García-Mili '26Liu '23Gilchrist '21
66% semantic matchBioinformatics
Zuguang Gu, Daniel Hübschmann·Genomics Proteomics & Bioinformatics·Jun 6, 2022· 311 citations