Reactome Database Kit

Query and analyze biological pathways from Reactome, a curated open-source pathway database with 2,800+ human pathways. This skill covers pathway search, enrichment analysis, pathway visualization, cross-species comparisons, and integrating pathway data with gene expression analysis.

When to Use This Skill

Choose Reactome Database Kit when you need to:

• Perform pathway enrichment analysis on gene lists from experiments
• Explore detailed molecular pathway diagrams and reaction steps
• Cross-reference genes with curated biological pathways and reactions
• Compare pathway conservation across species (human, mouse, rat, etc.)

Consider alternatives when:

• You need metabolic pathway-focused analysis (use KEGG)
• You need gene ontology term enrichment (use GO enrichment tools)
• You need drug-target-pathway relationships (use Open Targets)

Quick Start

Copy
pip install requests pandas matplotlib

Copy
import requests
import pandas as pd

BASE_URL = "https://reactome.org/ContentService"

# Search for pathways
response = requests.get(
    f"{BASE_URL}/search/query",
    params={"query": "apoptosis", "types": "Pathway", "species": "Homo sapiens"}
)
results = response.json()

for entry in results.get("results", [])[:5]:
    for item in entry.get("entries", []):
        print(f"{item['stId']}: {item['name']}")

Core Concepts

API Endpoints

Endpoint	Description	Example
/search/query	Full-text search	Search pathways by keyword
/data/pathway/{id}	Pathway details	Get pathway information
/data/pathway/{id}/containedEvents	Sub-pathways and reactions	Pathway hierarchy
/data/participants/{id}	Molecular participants	Genes/proteins in pathway
/analysis/identifiers	Enrichment analysis	Gene list analysis
/analysis/token/{token}	Analysis results	Retrieve enrichment results

Pathway Enrichment Analysis

Copy
import requests
import pandas as pd

def reactome_enrichment(gene_list, species="Homo sapiens"):
    """Perform pathway enrichment analysis via Reactome API."""
    # Submit gene list for analysis
    response = requests.post(
        "https://reactome.org/AnalysisService/identifiers/projection",
        headers={"Content-Type": "text/plain"},
        data="\n".join(gene_list),
        params={"interactors": "false", "pageSize": 50, "page": 1}
    )
    data = response.json()

    # Extract pathway results
    pathways = []
    for pathway in data.get("pathways", []):
        pathways.append({
            "pathway_id": pathway["stId"],
            "name": pathway["name"],
            "p_value": pathway["entities"]["pValue"],
            "fdr": pathway["entities"]["fdr"],
            "found": pathway["entities"]["found"],
            "total": pathway["entities"]["total"],
            "ratio": pathway["entities"]["ratio"]
        })

    df = pd.DataFrame(pathways)
    df = df.sort_values("fdr")

    print(f"Significant pathways (FDR < 0.05): "
          f"{(df['fdr'] < 0.05).sum()}/{len(df)}")

    return df

# Enrichment for cancer-related genes
genes = ["TP53", "BRCA1", "BRCA2", "EGFR", "KRAS", "MYC",
         "AKT1", "PIK3CA", "PTEN", "RB1", "CDK4", "BRAF"]

enriched = reactome_enrichment(genes)
print(enriched[["name", "fdr", "found", "total"]].head(15))

Configuration

Parameter	Description	Default
species	Target organism	"Homo sapiens"
interactors	Include protein-protein interactions	false
page_size	Results per page	20
sort_by	Sort enrichment results	"ENTITIES_FDR"
p_value_cutoff	Maximum p-value for significance	0.05
projection	Project to human orthologs	true

Best Practices

1. Use gene symbols, not Ensembl IDs, for the analysis API — Reactome's enrichment endpoint works best with HGNC gene symbols. Convert Ensembl IDs to symbols before submission. Mixed identifier types in a single submission reduce mapping coverage.

2. Apply FDR correction, not raw p-values — With 2,800+ pathways tested, raw p-values produce many false positives. Use the fdr field from results (Benjamini-Hochberg corrected) with a cutoff of 0.05 for reliable pathway identification.

3. Check pathway hierarchy for redundancy — Reactome pathways are hierarchical (e.g., "Apoptosis" contains "Intrinsic Pathway of Apoptosis"). Enrichment often flags both parent and child pathways. Use the hierarchy API to identify and collapse redundant results.

4. Use projection for non-human species — Set projection=true to map non-human genes to their human orthologs before enrichment. This leverages Reactome's extensive human pathway curation even when working with mouse, rat, or other model organism data.

5. Combine with expression data for context — Enrichment tells you which pathways are overrepresented, but not the direction of change. Overlay fold-change data onto enriched pathways using Reactome's expression analysis mode to distinguish upregulated vs downregulated pathways.

Common Issues

Gene list returns zero enriched pathways — Check that gene identifiers match Reactome's expected format (HGNC symbols for human). Test with known pathway genes first (e.g., TP53, BRCA1) to verify the API is working. Also check that the gene list isn't too small (<5 genes) for statistical enrichment.

Analysis token expires — Reactome analysis results are stored temporarily with a token. Tokens expire after a few hours. Save the full results immediately after analysis rather than relying on retrieving them later with the token.

Same pathway appears multiple times — Reactome distinguishes between species-specific pathway instances. Filter results by species with the species parameter to avoid seeing the same pathway for human, mouse, and other organisms. Also deduplicate by parent pathway ID to handle hierarchical redundancy.