Eight Practical Tips for Navigating the Reactome Pathway Knowledgebase

Researchers who need to make sense of genes, proteins and the molecular processes that link them now have a detailed practical guide to one of biology’s most widely used reference resources. In an article published in PLOS Computational Biology, a team from the Ontario Institute for Cancer Research and the NYU Grossman School of Medicine lays out eight quick tips for using the Reactome knowledgebase, a curated, freely accessible database that represents human molecular physiology as a connected set of biological processes. The guide, written by Nancy T. Li, Marija Milacic, Karen Rothfels and Peter D’Eustachio, is aimed at anyone who wants to move beyond treating pathway databases as black boxes and instead use them deliberately, reproducibly and to their full extent.

Reactome occupies a specific niche in the landscape of bioinformatics resources. Its central goal, as the authors describe it, is the systematic annotation of the functions of human proteins, providing standardized descriptions that support computer-based exploration of human biology and disease. The knowledgebase can be used in two complementary ways: as a searchable online textbook of human molecular physiology, and as a toolkit for interpreting high-throughput proteomic and other omic-scale data sets. Users can start from an individual gene, protein or reaction and navigate outward to the biological pathways in which that entity participates, or work in the opposite direction, from a broad process down to its molecular steps.

The scale of the resource is substantial. As of June 2026, Reactome contains annotations of 16,423 reactions involving 11,694 human proteins, out of an estimated 20,533 encoded by the human genome according to Ensembl, together with 2,188 small molecules. These annotations are backed by data drawn from 43,308 publications. The resource is actively curated and updated quarterly, which means that both the content and the pathway representations available to users change over time. That dynamism is one of the guide’s recurring themes: what a user downloads or queries today may differ, in meaningful ways, from what the same query returns a year later.

Understanding how Reactome represents biological knowledge is the foundation for using it well. The data model builds on classic metabolic maps, first implemented online by resources such as KEGG and EcoCyc/MetaCyc, and extends them into domains including gene expression, the cell cycle, differentiation and development, immune function, and diseases caused by infectious agents and by inherited or somatic mutations. At its core, a Reactome reaction is a single-step event in which input physical entities are transformed into output ones, enabled by proteins and protein complexes, and positively or negatively regulated by other entities. Transformations include covalent modifications, transport between subcellular locations, and the assembly and disassembly of multimolecular complexes. Because shared physical entities connect reactions, the output of one reaction can serve as the input, catalyst or regulator of the next, allowing users to trace continuous chains of causality within and between pathways.

The model also relies on careful links to community data resources, which serve as single sources of truth. Chemical entities are annotated as variants of canonical forms described by ChEBI and RHEA, distinguished from one another by subcellular location. Proteins are annotated as variants of canonical UniProt forms, distinguished by posttranslational modifications, subcellular location and cell type. Gene Ontology terms identify molecular functions, roles in multistep processes and locations. This referencing system means that a user can find all annotations of a canonical entity and its variants and navigate to the definitive description maintained by the external resource. Notably, Reactome does not aim to be a comprehensive repository of everything ever published; its annotations are deliberately terse and current, synthesizing evidence into a structured consensus view rather than cataloguing every observation or tracing how current views arose.

The first tip in the guide is to use Reactome as a searchable online textbook. The primary interface at www.reactome.org offers a hierarchy of superpathways, such as metabolism and DNA replication, that can be unfurled to reveal subpathways and, ultimately, single-step reactions. Selecting any event in the hierarchy displays its details in a standard human-readable format. A user can navigate, for example, from Metabolism to Carbohydrate metabolism to Glucose metabolism to Glycolysis and finally to a specific reaction such as the cleavage of fructose-1,6-bisphosphate by the aldolase tetramer. Alternatively, direct searches for a gene, protein, chemical or event return the reactions and pathways in which that entity participates, along with related disease processes and inferred events involving model-organism orthologs.

The second tip addresses a common frustration: not knowing which terms to search for in the first place. For questions broader than a single gene, protein or pathway, the authors point to the React-to-Me chatbot, which accepts questions in English and more than twenty non-English languages and retrieves relevant curated content from Reactome. A question such as “What is the role of GATA6 in the pancreas?” draws on information from several areas of the knowledgebase rather than a single pathway, and the chatbot presents results as a combination of text, pathway diagrams, illustrations and links to related resources. The team notes that more detail about the chatbot is available in a separate preprint.

The third and fourth tips concern data analysis. Reactome’s website and application programming interface provide tools for qualitative and quantitative gene set enrichment and for cross-species and cross-tissue pathway comparison, with results explorable in the context of the underlying pathway hierarchy. Specialized tools extend these capabilities: ReactomeFIViz and ReactomeIDG support analysis of interaction and perturbation effects among genes of interest, based on their annotated Reactome roles and coregulation data imported from external sources. For researchers working with non-human organisms, Reactome offers computationally inferred pathway representations for fourteen model organisms. When a model organism has orthologs of the proteins in a human reaction, the corresponding reaction is inferred by sequence similarity and assembled into model-organism pathways, which can be compared with the curated human set through a species comparison tool. The authors stress an important caveat: these pathways are computationally inferred, not manually curated, and users should distinguish inferred events from experimentally supported human annotations when interpreting results.

The remaining tips address rigor, support and community. Because Reactome updates quarterly and those changes can affect pathway analysis results, the guide advises recording the release date and version number alongside other analysis parameters, preserving test results and successive data versions, and rerunning analyses to distinguish changes arising from the data from changes arising from the resource itself. A history-tracking feature called UpdateTracker helps identify whether relevant annotations were modified between releases. Citing the relevant Reactome publications and acknowledging the version used is described as essential, both for reproducibility and to justify continued funding of the project, which is supported by grants from the National Institutes of Health, core funding from the European Bioinformatics Institute, Open Targets and Connected Minds Prototyping at York University. Users seeking help can turn to a published tutorial, online documentation including a data model glossary, video tutorials and webinars, a help desk reachable at [email protected], and even individual consultations for complex data sets or publication figures.

The final two tips invite participation and reuse. Reactome is a community-curated resource, and researchers who find annotations out of date or incomplete can review existing pathway content or collaborate with curators to add or extend pathways, receiving authorship or reviewership credit, ORCID attribution through a claiming feature, and, for every pathway built with outside experts, a citable DOI. The resource’s content and software are also freely available for reuse on an open-source basis: data files and reports can be downloaded, and the open-access library of curated pathway diagrams and icons can be used to create publication-ready illustrations. A machine-facing API and related tools at reactome.org/dev allow programmatic access, so that a researcher building a bioinformatics workflow can retrieve pathway data automatically rather than downloading individual pathways by hand. Taken together, the authors argue, these practices, from understanding the pathway hierarchy to engaging with the community, help ensure that pathway-based analyses are accurate, reproducible and aligned with current biological knowledge.

Subject of Research: Practical guidance for using the Reactome curated pathway knowledgebase in biological research

Article Title: Eight quick tips for using the reactome knowledgebase

Article References: Li, N. T., Milacic, M., Rothfels, K., & D’Eustachio, P. (2026). Eight quick tips for using the reactome knowledgebase. PLOS Computational Biology, 22(10), e1014891. https://doi.org/10.1371/journal.pcbi.1014891

Image Credits: AI Generated

DOI: 10.1371/journal.pcbi.1014891

Keywords: Reactome, bioinformatics, pathway analysis, knowledgebase, data curation, gene set enrichment, model organisms, data model, API, reproducibility, proteomics, open science