HolobiomicsLab Compute Relative Betweenness Centrality for Metabolites in Pathway Networks
Summary
Quantify the topological importance and regulatory influence of metabolites within a pathway-metabolite bipartite network by computing betweenness centrality and normalizing to Relative Betweenness Centrality (RBC) scores in the [0,1] range. This identifies hub metabolites that bridge multiple pathways.
When to use
Apply this skill when you have (1) a set of input metabolites (e.g., from differential metabolomics analysis), (2) a pathway-metabolite mapping file defining bipartite edges, and (3) you need to rank metabolites by their topological role in the network to identify regulatory bottlenecks or central hubs that warrant further mechanistic investigation.
When NOT to use
- Input metabolites are already ranked or scored by another method (e.g., fold-change, p-value): betweenness centrality measures network topology, not statistical significance
- Network is not bipartite or metabolite-pathway structure is unknown: centrality interpretation depends on valid bipartite structure
- Sample size is very small (< 5 metabolites in input set) or network is sparse: betweenness scores may be uninformative with insufficient connectivity
Inputs
- pathway-metabolite bipartite network (edge list or adjacency matrix)
- input metabolite list (character vector of KEGG IDs or metabolite identifiers)
- optional: metabolite-level annotations (names, pathway assignments)
Outputs
- data.frame with columns: metabolite identifier, raw betweenness centrality, Relative Betweenness Centrality (RBC) score [0,1]
- Relative Betweenness Centrality (RBC) plot (ranked barplot or scatter visualization)
- optional: igraph object representing the filtered pathway-metabolite network
How to apply
Load the pathway-metabolite network and input metabolite list into R. Construct a bipartite igraph object with metabolites and pathways as nodes and membership relationships as edges. Filter the graph to retain only metabolites in your input set. Compute raw betweenness centrality for each metabolite node using igraph::betweenness(). Normalize betweenness values to the [0,1] range by dividing by the graph's maximum possible centrality (or scaling to the observed maximum), producing Relative Betweenness Centrality (RBC) scores. Generate a data.frame with metabolite identifiers, RBC scores, and optional pathway annotations. Visualize the RBC distribution using a ranked barplot or scatter plot to highlight which input metabolites occupy central topological positions.
Related tools
- igraph (Construct bipartite graph, compute betweenness centrality for metabolite and pathway nodes) — https://igraph.org
- KEGGREST (Retrieve pathway-to-metabolite mappings from KEGG database) — https://bioconductor.org/packages/KEGGREST
- enrichmet (Wrapper function that integrates graph construction, centrality computation, and RBC visualization into a single workflow) — https://github.com/biodatalab/enrichmet
- R (Execution environment for igraph and data manipulation)
Examples
results <- enrichmet(inputMetabolites = inputMetabolites, PathwayVsMetabolites = PathwayVsMetabolites, example_data = example_data, analysis_type = c('centrality'), top_n = 15)
Evaluation signals
- RBC values fall within [0, 1] range and sum of normalized scores across all metabolites is meaningful relative to network density
- Metabolites with high RBC appear in multiple pathways in the input mapping; low-RBC metabolites are peripheral or pathway-specific
- RBC plot is sorted in descending order; visual inspection confirms hub metabolites (e.g., glucose, ATP, CoA) rank highest in typical metabolic networks
- Betweenness centrality and RBC rank order remain stable when recomputing on the same network, confirming reproducibility
- No metabolites have undefined or infinite centrality scores; all input metabolites present in output data.frame with valid numeric RBC
Limitations
- Betweenness centrality is computationally expensive (O(n²) or O(n³) depending on algorithm); large networks (>10,000 nodes) may require approximation
- Centrality scores depend critically on network completeness and accuracy; missing pathway-metabolite edges or incorrect mapping undermine interpretation
- Isolated metabolites (degree 0) have zero betweenness; filtering or separate handling may be required for complete input sets
- RBC normalization scheme affects absolute scores; different normalization methods (divide by max, by theoretical maximum, by graph diameter) yield non-comparable numeric values across different studies
- Bipartite network topology may obscure direct metabolite-metabolite interactions (which are not represented as edges); consider supplementary analysis with metabolite-only or pathway-only projections
Evidence
- [intro] task_003_finding_rbc_computation: "enrichmet computes betweenness centrality for metabolites and produces a Relative Betweenness Centrality (RBC) plot that displays RBC values on the x-axis to highlight the topological importance and"
- [intro] task_003_workflow_step_1: "Load the pathway-metabolite network data and KrasG12D metabolite input list into R. Construct the pathway-metabolite bipartite graph using igraph, with nodes representing both pathways and"
- [intro] task_003_workflow_step_2: "Filter the network to retain only metabolites in the KrasG12D input set. Compute betweenness centrality for each metabolite node using igraph's betweenness centrality function. Normalize betweenness"
- [intro] task_003_workflow_step_3: "Generate a data.frame containing metabolite identifiers, RBC scores, and optional pathway annotations. Create the RBC plot visualization (e.g., barplot or ranked scatter) showing RBC values for all"
- [intro] enrichmet_igraph_integration: "enrichmet integrates fgsea for fast MetSEA, igraph for topology-based metrics, and curated KEGG data for enrichment using Fisher's Exact Test—all accessible via a single function call"