The Reactome Pathway Knowledgebase: A Comprehensive Veterinary Systems Biology Resource

Introduction

The Reactome Pathway Knowledgebase represents a freely accessible, manually curated, and peer-reviewed repository of biological pathways that has evolved into a cornerstone resource for molecular systems biology [1, 2]. Initially developed primarily for human biology, Reactome has expanded its utility to encompass model organisms, plants, and veterinary species through robust orthology inference methods and dedicated curation projects such as Plant Reactome [3, 4]. For veterinary researchers and diagnosticians, Reactome offers a systematic framework for interpreting molecular data from animal tissues, pathogens, and host-pathogen interactions, thereby supporting hypothesis generation, biomarker discovery, and therapeutic target identification [5].

The knowledgebase operates on a reaction-centric data model where each molecular event is defined by its participants (inputs, outputs, catalysts), cellular location, and regulatory modifiers [6]. These reactions are assembled into chains that form pathways, which are further organized into hierarchical categories such as metabolism, signal transduction, cell cycle, and immune system. As of the most recent releases, Reactome contains over 10,000 manually curated reactions involving more than 10,000 proteins, with extensive annotations of post-translational modifications, disease variants, and drug-target interactions [2, 7].

Data Model and Curation

The core unit of Reactome is the Reaction, which represents a molecular transformation such as a binding event, phosphorylation, translocation, or complex dissociation. Each reaction is linked to specific physical entities (proteins, small molecules, complexes) and is constrained by a cellular compartment (e.g., cytosol, mitochondrion, nucleus) [8]. Reactions are connected through shared entities to form pathways, and pathways are grouped into pathway hierarchies.

Curation follows a rigorous expert-author model. Domain specialists collaborate with Reactome editorial staff to create detailed reaction descriptions supported by primary literature. Each entry is peer-reviewed by a second expert before release [1, 9]. The quality control pipeline includes automated checks for data consistency, logical completeness, and cross-referencing with external databases such as UniProt, PubMed, and the Gene Ontology [10, 11]. A recent development includes automated monitoring of retracted publications to maintain data integrity [7].

The Reactome data model has been extended to support annotation of disease processes, including those caused by infectious agents and genetic mutations [5]. This extension is particularly relevant for veterinary virology, as it enables detailed representation of viral life cycles and host immune responses within a consistent framework. For example, Reactome now includes curated pathways for coronavirus replication and host cell manipulation, which can be adapted to veterinary coronaviruses through orthology mapping [12, 5].

Orthology Inference and Cross-Species Applicability

A defining feature of Reactome for the veterinary community is its ability to project human pathways onto other species through electronic orthology inference. At each quarterly release, Reactome uses ortholog predictions from resources such as Ensembl Compara and InParanoid to create species-specific reaction predictions for over 80 species, including mammals, birds, fish, amphibians, and invertebrates [4, 6]. This projection process assigns human proteins to their one-to-one or one-to-many orthologs in the target species, allowing the human pathway model to serve as a template for molecular events in animals.

For veterinary species, this orthology-based projection provides a powerful starting point for interpreting high-throughput data. For example, a researcher studying the mammary gland transcriptome of dairy cattle affected by mastitis can map bovine gene expression data onto Reactome's human pathways; the underlying molecular logic of immune signaling and metabolism is largely conserved, enabling functional interpretation of differentially expressed genes. However, caution is required because gene duplication, functional divergence, and species-specific physiology (e.g., ruminant digestion, avian uric acid metabolism) may not be fully captured.

To address these limitations, dedicated curation projects such as Plant Reactome have been established. Plant Reactome uses rice (Oryza sativa) as a reference species for manual curation and then projects pathways to over 120 plant species via orthology [3, 13]. This resource is directly relevant to veterinary toxicology and nutrition, as many animal feeds contain plant metabolites, and plant pathways can help model detoxification mechanisms in herbivores.

Applications in Veterinary Research and Diagnostics

Reactome is widely used for pathway enrichment analysis of transcriptomic and proteomic datasets. The ReactomeGSA platform (Gene Set Analysis) allows users to upload gene expression data from any species, provided appropriate gene identifiers are used, and performs over-representation analysis or gene set enrichment analysis against Reactome pathways [14]. The platform supports simultaneous analysis of multiple datasets, enabling comparative studies across tissues, treatments, or species. For veterinary diagnostics, this can be applied to identify pathway dysregulation in infectious diseases, metabolic disorders, or neoplasia.

An example application is the elucidation of host-pathogen interactions. Reactome includes curated pathways for the life cycles of several human viruses and their interaction with host signaling networks [5]. By orthology inference, these pathways can be adapted to veterinary viruses such as feline coronavirus, canine distemper virus, or avian influenza virus. The Reactome data model allows annotation of viral proteins as physical entities that engage host complexes, enabling mechanistic modeling of infection and immune evasion.

The Reactome-IDG (Illuminating the Druggable Genome) portal further extends utility by contextualizing understudied proteins within pathways [15, 16]. This is valuable for veterinary pharmacology where drug targets in animals may be poorly characterized. The portal uses machine learning to predict functional interactions between dark proteins and pathway-annotated proteins, providing a basis for hypothesis-driven experimental validation.

For metabolomics, Reactome provides comprehensive metabolite-to-pathway annotations [17, 18]. Machine learning models trained on Reactome data can predict pathway involvement for detected metabolites, significantly improving pathway enrichment analysis of metabolomics datasets [19]. This is particularly useful in veterinary toxicology, where identification of disrupted metabolic pathways helps assess the impact of environmental toxins or feed contaminants.

Visualization and Analysis Tools

Reactome offers a suite of visualization tools designed to support interactive exploration of pathway data. The PathwaBrowser diagram viewer is built on a layered HTML5 canvas architecture with space-partitioning data structures that achieve sub-second rendering for 97% of diagrams [20]. Entity-level views (ELVs) provide detailed molecular schematics including chemical structures and animated protein models [7]. A "compare mode" allows side-by-side visualization of normal and disease states, which can be used to depict wild-type versus variant phenotypes in veterinary genetic diseases.

The ReacFoam tool provides a global pathway overview using a 2D foam-like layout, enabling users to identify broad functional categories and their interconnections [7]. For programmatic access, Reactome's ContentService REST API and graph database (Neo4j) allow complex queries that traverse the highly interconnected data model with 93% reduction in query time compared to relational storage [9].

ReactomeFIViz, a Cytoscape app, integrates Reactome pathways with genome-wide functional interaction networks, enabling visualization of drug-target interactions in the context of pathways and networks [21]. This is useful for veterinary precision medicine, where understanding how a drug's mechanism overlaps with a patient's genetic background can inform treatment decisions.

Integration with Other Resources and FAIR Compliance

Reactome is deeply integrated with other biomedical data resources. It provides cross-references to UniProt, ChEBI, PubMed, GO, COSMIC, PRIDE, and many others [1, 2, 10]. The Pathways2GO converter transforms Reactome reactions into Gene Ontology Causal Activity Models (GO-CAMs), enabling interoperability between these two major pathway representations [10].

Reactome has achieved CoreTrustSeal certification and is designated as a Global Core Biodata Resource and ELIXIR core resource, underscoring its commitment to FAIR (Findable, Accessible, Interoperable, Reusable) data principles [7]. All data are available under Creative Commons Attribution 4.0 license, and the software is open source [22].

The React-to-Me conversational interface employs retrieval-augmented generation (RAG) to provide natural language access to Reactome content while maintaining source traceability [23]. Users can ask complex questions and receive answers grounded in curated pathway data, which may assist veterinary clinicians in interpreting molecular findings without deep bioinformatics expertise.

Limitations and Considerations for Veterinary Use

Despite its breadth, Reactome has limitations that warrant consideration. The majority of curational effort focuses on human biology; orthology-based projections to veterinary species are inherently incomplete and may lack species-specific interactions. For example, immune system pathways in ruminants exhibit unique features such as γδ T cell expansion that are not represented in the human reference. Similarly, avian-specific metabolic pathways (e.g., uric acid excretion) and reproductive processes are absent.

The use of Reactome for veterinary species requires careful mapping of gene identifiers. While Reactome accepts multiple identifier types (Ensembl, Entrez, UniProt), less well-annotated genomes may have incomplete coverage. Additionally, the database primarily supports protein-coding genes; non-coding RNAs and repeat elements are not represented, limiting its application to certain regulatory studies.

The quality of orthology inference varies across gene families. Multi-copy gene families often have ambiguous orthology assignments, leading to false positives or negatives in pathway projections. Users should validate predicted pathway involvement through literature review or experimental evidence.

Future Directions and Emerging Capabilities

Recent developments indicate several avenues for enhanced veterinary utility. The integration of large language models (LLMs) into the curation workflow promises to accelerate annotation of new genes and pathways, potentially including more non-human species [24]. The Reactome Knowledgebase 2026 release introduces community-driven tutorials and an open icon library, lowering barriers for researchers to create educational materials tailored to veterinary curricula [7].

Predictive modeling for metabolite-pathway involvement has expanded to encompass KEGG, MetaCyc, and Reactome datasets simultaneously, using graph convolutional neural networks with chemical structure standardization achieved via InChI canonicalization [18]. This multi-knowledgebase approach increases the number of pathways available for prediction and improves generalization to novel compounds, which is critical for veterinary toxicology studies where unique xenobiotics may be encountered.

The development of PathwayMatcher enables proteoform-centric network construction, accounting for isoform and post-translational modification states of proteins [25]. This higher granularity may allow veterinary researchers to examine how species-specific splice variants or phosphorylation sites affect pathway activity in disease.

Conclusion

The Reactome Pathway Knowledgebase serves as a foundational resource for systems-level interpretation of molecular data across species. Through expert curation, orthology projection, and a comprehensive suite of analysis tools, it enables veterinary researchers to place experimental observations in the context of conserved biological processes. While limitations exist for non-human species, ongoing development of plant pathways, dark protein functional prediction, and AI-assisted curation continues to expand its relevance. For veterinary clinicians and diagnosticians seeking mechanistic understanding of disease and treatment, Reactome provides an indispensable framework.

flowchart TD
    A[High-Throughput Data from Veterinary Sample], > B{Data Type}
    B, > C[Transcriptomics]
    B, > D[Proteomics]
    B, > E[Metabolomics]
    C, > F[Gene Identifier Mapping to Reactome]
    D, > F
    E, > G[Metabolite-to-Pathway Prediction ML Model]
    G, > H[ReactomeGSA]
    F, > H
    H, > I[Pathway Enrichment Analysis]
    I, > J[Orthology-Based Pathway Projection]
    J, > K[Species-Specific Interpretation]
    K, > L[Hypothesis Generation & Experimental Validation]
    L, > M[Clinical Application / Therapeutic Target Identification]

References

[1] A. Fabregat, S. Jupe, L. Matthews et al. "The Reactome pathway knowledgebase." Nucleic Acids Res. 2013. https://www.semanticscholar.org/paper/d8bd861b292233072aa40ab6bc029ac94d7f30af

[2] Marija Milacic, Deidre Beavers, P. Conley et al. "The Reactome Pathway Knowledgebase 2024." Nucleic Acids Research. 2023. https://www.semanticscholar.org/paper/646303a7bf1985f799a329a7666a7eb7f89898b5

[3] Parul Gupta, J. Elser, Elizabeth Hooks et al. "Plant Reactome Knowledgebase: empowering plant pathway exploration and OMICS data analysis." Nucleic Acids Research. 2023. https://www.semanticscholar.org/paper/aa221f53d09eb5feee9f14ac927ec08d31427af4

[4] L. Matthews, Gopal R. Gopinath, M. Gillespie et al. "Reactome knowledgebase of human biological pathways and processes." Nucleic Acids Res. 2008. https://www.semanticscholar.org/paper/e904839fed3aa5f5937a28e1b1a6f9f3fbe1fa09

[5] L. Matthews, Justin Cook, Ralf Stephan et al. "Advancing curation of viral life cycles, host interactions, and therapeutics in Reactome." Journal of Virology. 2025. https://www.semanticscholar.org/paper/970000e0b5f8ba3fb1465a7e25ea9d25f161cf5b

[6] L. Stein, Peterd ' Eustachio, Gopal Gopinathrao et al. "Reactome: a knowledgebase of biological pathways." Nucleic Acids Res. 2004. https://www.semanticscholar.org/paper/4825530840556d105e5a7a10ec3b5d7d0aa95ec5

[7] E. Ragueneau, Chuqiao Gong, Pierre Sinquin et al. "The Reactome Knowledgebase 2026." Nucleic Acids Research. 2025. https://www.semanticscholar.org/paper/aeccaeac9fc0563269c8a64fa5b0eb05aa7d2e4f

[8] P. D’Eustachio, David Croft, B. Bono et al. "Reactome - a knowledgebase of human biological pathways." Journal. 2007. https://www.semanticscholar.org/paper/2932ae9857a621515ea5558846ce4bc65d72f547

[9] A. Fabregat, Florian Korninger, Guilherme Viteri et al. "Reactome graph database: Efficient access to complex pathway data." PLoS Comput. Biol. 2018. https://www.semanticscholar.org/paper/e4b52a1a00e9db941326fc857b95245cbfb60bdc

[10] Benjamin M. Good, K. V. Auken, D. Hill et al. "Reactome and the Gene Ontology: digital convergence of data resources." Bioinform. 2021. https://www.semanticscholar.org/paper/968f9286d57fe48ae9335d061516361709f7c58f

[11] Wang LL, Thomas Hayman G, Smith JR et al. "Predicting instances of pathway ontology classes for pathway integration." J Biomed Semantics. 2019. https://pubmed.ncbi.nlm.nih.gov/31196182/

[12] M. Gillespie, B. Jassal, Ralf Stephan et al. "The reactome pathway knowledgebase 2022." Nucleic Acids Res. 2021. https://www.semanticscholar.org/paper/dfc3df1eea66af990cd1731bb21111f2c2088467

[13] Sushma Naithani, Parul Gupta, Justin Preece et al. "Plant Reactome: a knowledgebase and resource for comparative pathway analysis." Nucleic Acids Res. 2019. https://www.semanticscholar.org/paper/dd1270c819abcb9fca19b98c1a1463c7ad465aac

[14] A. Grentner, E. Ragueneau, Chuqiao Gong et al. "ReactomeGSA: new features to simplify public data reuse." Bioinform. 2024. https://www.semanticscholar.org/paper/110e0755813c35bf9e4c17e7bbe9309ab24c2342

[15] Deidre Beavers, Timothy Brunson, Nasim Sanati et al. "Illuminate the Functions of Dark Proteins using the Reactome-IDG Web Portal." Current Protocols. 2023. https://www.semanticscholar.org/paper/edc8cd49ef194b63170b063052f2232c60d1a68c

[16] Timothy Brunson, Nasim Sanati, L. Matthews et al. "Illuminating Dark Proteins using Reactome Pathways." bioRxiv. 2023. https://www.semanticscholar.org/paper/510d50205ceca6ee909fb22a85d8663c06340611

[17] Erik D. Huckvale, H. Moseley. "Predicting the Pathway Involvement of Compounds Annotated in the Reactome Knowledgebase." Metabolites. 2025. https://www.semanticscholar.org/paper/a2b7552130f8410cdf3cf37dfbdb92f4f8dd72f3

[18] Erik D. Huckvale, H. Moseley. "Chemical Structure Representation Standardization Is Needed to Generalize Metabolite-Pathway Involvement Prediction Across KEGG, Reactome, and MetaCyc Knowledgebases." Metabolites. 2026. https://www.semanticscholar.org/paper/4e676c90123f86982565cd007a4a55bd55980826

[19] Erik D. Huckvale, P. T. Thompson, R. Flight et al. "High-Quality Predicted Pathway Annotations Greatly Improve Pathway Enrichment Analysis of Metabolomics Datasets." bioRxiv. 2025. https://www.semanticscholar.org/paper/183491a52332d3b2957cda0b7d7ccd65fd76c7d7

[20] A. Fabregat, K. Sidiropoulos, Guilherme Viteri et al. "Reactome diagram viewer: data structures and strategies to boost performance." Bioinform. 2017. https://www.semanticscholar.org/paper/7fde8abd1df28b502982d12b7361b18e5f796a7d

[21] A. Blucher, S. McWeeney, L. Stein et al. "Visualization of drug target interactions in the contexts of pathways and networks with ReactomeFIViz." F1000Research. 2019. https://www.semanticscholar.org/paper/660146ca6574e6818c01c801a2419cd0607b976e

[22] S. McKay, Joel Weiser. "Installing a Local Copy of the Reactome Web Site and Knowledgebase." Current Protocols in Bioinformatics. 2015. https://www.semanticscholar.org/paper/4188a91f2f7aa48c3d9304de9f0f63926256f4d4

[23] Helia Mohammadi, Fatemeh Almodaresi, Gregory F J Hogue et al. "React-to-Me: A Conversational Interface for Interactive Exploration of the Reactome Pathway Knowledgebase." bioRxiv. 2025. https://www.semanticscholar.org/paper/2d7a9b91ffc68c4838f51498ee1031de48b2112e

[24] Guanming Wu, L. Matthews, N. Boyer et al. "Application of Large Language Models for Annotating Genes into Reactome Pathways." bioRxiv. 2025. https://www.semanticscholar.org/paper/d0f35111174f3dcebf98ab9297d163ad99e784cb

[25] Luis Francisco Hernández Sánchez, Bram Burger, Carlos Horro et al. "PathwayMatcher: proteoform-centric network construction enables fine-granularity multi-omics pathway mapping." bioRxiv. 2018. https://www.semanticscholar.org/paper/ed4269253b4a0a36f500197184205f7306b419b2

[26] Parul Gupta, Sushma Naithani, Justin Preece et al. "Plant Reactome and PubChem: The Plant Pathway and (Bio)Chemical Entity Knowledgebases." Methods in Molecular Biology. 2022. https://www.semanticscholar.org/paper/07683250b6087cde0ad39cc0634145988c22f9de

[27] Thawfeek M. Varusai, S. Jupe, Cristoffer Sevilla et al. "Using Reactome to build an autophagy mechanism knowledgebase." Autophagy. 2020. https://www.semanticscholar.org/paper/46d254c0ea59c63c0f7dbf5511040138a79261c2

[28] S. Racunas, N. Shah, N. Fedoroff. "A case study in pathway knowledgebase verification." BMC Bioinformatics. 2006. https://www.semanticscholar.org/paper/5c0738d3e66773f237b2ee7795368be33a194bea

[29] B. Bono. "The Breadth and Depth of BioMedical Molecular Networks: The Reactome Perspective." Journal. 2009. https://www.semanticscholar.org/paper/b460c6ff6d4a64b8a4b2f3382b8c29fe35aafb82

[30] Adam J. Wright, M. Orlic-Milacic, K. Rothfels et al. "Evaluating the predictive accuracy of curated biological pathways in a public knowledgebase." Database J. Biol. Databases Curation. 2022. https://www.semanticscholar.org/paper/48d62bcb88714d0f47a1b04ebd5977769772805b

[31] L. Matthews, P. D’Eustachio, David Croft et al. "An Introduction to the Reactome Knowledgebase of Human Biological Pathways and Processes." Journal. 2007. https://www.semanticscholar.org/paper/cb38cacc6012a8f0669f3603f568739359c7f405

[32] A. Fabregat, K. Sidiropoulos, Guilherme Viteri et al. "Reactome: A curated knowledgebase of biomolecular pathways." Journal. 2016. https://www.semanticscholar.org/paper/01d76b31104c67057480254d08bc0c3bfabca4b1

[33] S. Jupe, J. Akkerman, N. Soranzo et al. "Reactome - a curated knowledgebase of biological pathways: megakaryocytes and platelets." Journal of Thrombosis and Haemostasis. 2012. https://www.semanticscholar.org/paper/3ee48be12e9672fdccb0883c60060bf51aa23990

[34] N. Ndegwa, R. Côté, David Ovelleiro et al. "Critical amino acid residues in proteins: a BioMart integration of Reactome protein annotations with PRIDE mass spectrometry data and COSMIC somatic mutations." Database J. Biol. Databases Curation. 2011. https://www.semanticscholar.org/paper/7d61123306e0f31d70afe9fe1cf7644957e8d51e

[35] Nguyen H, Pham VD, Nguyen H et al. "CCPA: cloud-based, self-learning modules for consensus pathway analysis using GO, KEGG and Reactome." Brief Bioinform. 2024. https://pubmed.ncbi.nlm.nih.gov/39041916/