Skip to contents

PPI Network & Enrichment Analysis

ppi-enrichment.Rmd

This vignette demonstrates how to perform protein-protein interaction (PPI) network analysis and GO/KEGG pathway enrichment using the unitcm package. These functions call the public STRING and Enrichr APIs directly — no UniTCM API key is required.

Step 1: Define a Gene List

Start with a list of target genes — for example, the integrated targets from a network pharmacology analysis of a TCM formula:

genes <- c(
  "TP53", "BRCA1", "EGFR", "VEGFA", "TNF",
  "MYC", "AKT1", "IL6", "JUN", "MAPK3",
  "PTEN", "NFKB1", "STAT3", "HIF1A", "CCND1",
  "BCL2", "CASP3", "CTNNB1", "ERBB2", "FOS"
)
length(genes)

Step 2: Query the STRING PPI Database 

ppi <- query_string_ppi(genes, species = 9606, score_threshold = 400)
nrow(ppi)
head(ppi)

Step 3: Build the PPI Network 

g <- build_ppi_network(ppi, genes)
cat("Nodes:", vcount(g), "\n")
cat("Edges:", ecount(g), "\n")

Step 4: Identify Hub Genes

Hub genes are the most highly connected nodes in the network, ranked by degree centrality:

hubs <- identify_hub_genes(g, top_n = 10)
hubs

# Hub genes as a data frame
hub_df <- data.frame(
  Gene = names(hubs),
  Degree_Centrality = as.numeric(hubs)
)
hub_df

Step 5: Community Detection (Louvain Clustering)

Partition the PPI network into functional modules:

clusters <- louvain_cluster(g)
lengths(clusters)

# View members of the largest cluster
largest_idx <- which.max(lengths(clusters))
cat("Largest cluster (", lengths(clusters)[largest_idx], "genes):\n",
    paste(clusters[[largest_idx]], collapse = ", "))

Step 6: GO & KEGG Enrichment via Enrichr

Submit the gene list to Enrichr for functional enrichment analysis:

enrich <- enrichr_enrichment(genes, top_n = 10)
names(enrich)

KEGG Pathways 

kegg <- enrich[["KEGG_2021_Human"]]
kegg

GO Biological Process 

go_bp <- enrich[["GO_Biological_Process_2023"]]
go_bp

WikiPathways 

wp <- enrich[["WikiPathway_2021_Human"]]
wp

Full Workflow Summary

Putting it all together in a compact pipeline:

# 1. Query PPI
ppi <- query_string_ppi(genes)

# 2. Build network
g <- build_ppi_network(ppi, genes)

# 3. Hub genes
hubs <- identify_hub_genes(g, top_n = 10)

# 4. Functional modules
clusters <- louvain_cluster(g)

# 5. Pathway enrichment
enrich <- enrichr_enrichment(genes)

# 6. Report
cat("\n=== SUMMARY ===\n")
cat("Input genes:", length(genes), "\n")
cat("PPI edges:", nrow(ppi), "\n")
cat("Network nodes:", vcount(g), "  edges:", ecount(g), "\n")
cat("Top hub:", names(hubs)[1], " (degree centrality =", round(hubs[1], 4), ")\n")
cat("Communities:", length(clusters), "\n")
cat("Enriched KEGG pathways:", nrow(enrich[["KEGG_2021_Human"]]), "\n")

Notes

  • Species parameter: Use NCBI taxonomy IDs — 9606 for human, 10090 for mouse, 10116 for rat.
  • Score threshold: STRING combined score (0–1000). Values ≥ 700 represent high confidence, ≥ 400 medium confidence.
  • Enrichr libraries: The default set covers GO (BP, MF, CC), KEGG, and WikiPathways. See the Enrichr website for the full list of available gene-set libraries.
  • Rate limits: Both STRING and Enrichr are free, public APIs. The functions include built-in throttling and retries, but avoid sending large numbers of requests in rapid succession.