| name | bioconductor-visse |
| description | This package enables the interpretation and analysis of results from a gene set enrichment analysis using network-based and text-mining approaches. Most enrichment analyses result in large lists of significant gene sets that are difficult to interpret. Tools in this package help build a similarity-based network of significant gene sets from a gene set enrichment analysis that can then be investigated for their biological function using text-mining approaches. |
vissE
Workflows
Standard Workflow
Summarize and interpret a list of significant gene sets from an enrichment analysis using network clustering, text-mining, and gene/PPI visualization.
library(msigdb)
library(GSEABase)
library(vissE)
library(igraph)
library(ggplot2)
library(patchwork)
msigdb_hs = getMsigdb()
msigdb_hs = subsetCollection(msigdb_hs, c('h', 'c2', 'c5'))
set.seed(360)
geneset_res = sample(sapply(msigdb_hs, setName), 2500)
geneset_gsc = msigdb_hs[geneset_res]
gs_ovlap = computeMsigOverlap(geneset_gsc, thresh = 0.25)
gs_ovnet = computeMsigNetwork(gs_ovlap, msigdb_hs)
geneset_stats = rnorm(2500)
names(geneset_stats) = geneset_res
grps = findMsigClusters(gs_ovnet, genesetStat = geneset_stats, alg = cluster_walktrap, minSize = 5)
p1 = plotMsigWordcloud(msigdb_hs, grps[1:6], type = 'Name')
genes = unique(unlist(geneIds(geneset_gsc)))
gene_stats = rnorm(length(genes))
names(gene_stats) = genes
p3 = plotGeneStats(gene_stats, msigdb_hs, grps[1:6]) + geom_hline(yintercept = 0, colour = 2, lty = 2)
ppi = getIMEX('hs', inferred = TRUE)
p4 = plotMsigPPI(ppi, msigdb_hs, grps[1:6], geneStat = gene_stats, threshStatistic = 0.2, threshConfidence = 0.2)
p2 = plotMsigNetwork(gs_ovnet, markGroups = grps[1:6], genesetStat = geneset_stats)
p1 + p2 + p3 + p4 + plot_layout(2, 2)
Input: A GeneSetCollection object containing significant gene sets. Output: A combined paneled plot summarizing the gene-set clusters, text-mining word clouds, gene statistics, and protein-protein interactions.
When to Use
- To summarize large lists of significant gene sets from enrichment analyses (e.g., from
limma::fry, singscore, or GSEA) using network-based clustering.
- To perform text-mining on gene-set names or short descriptions using frequency analysis (adjusted with inverse document frequency) to identify recurring biological themes.
- To visualize gene-level statistics (using
plotGeneStats) and protein-protein interactions (using plotMsigPPI) within identified gene-set clusters.
When NOT to Use
- For performing the initial differential expression or gene-set enrichment calculation itself; use packages like
limma or singscore first.
- When you do not have gene-set definitions or a
GeneSetCollection object; vissE requires these to compute overlaps.
Data Requirements
- A
GeneSetCollection object (e.g., from GSEABase or msigdb) representing the significant gene sets.
- A named numeric vector of gene-set statistics (e.g., p-values or FDRs) where names match the gene-set names.
- A named numeric vector of gene-level statistics (e.g., log fold-changes) where names match gene identifiers (e.g., Gene Symbols).
Key Parameters
- thresh (0.25): Threshold for Jaccard index or overlap coefficient in
computeMsigOverlap.
- alg (
cluster_walktrap): Graph clustering algorithm from igraph used in findMsigClusters.
- minSize (5): Minimum cluster size to retain in
findMsigClusters.
- type ('Name'): Type of text-mining source ('Name' or 'Short') in
plotMsigWordcloud.
- threshStatistic (0.2): Minimum gene-level statistic threshold for filtering nodes in
plotMsigPPI.
- threshConfidence (0.2): Minimum confidence score for filtering PPI edges in
plotMsigPPI.
Best Practices
- Filter MSigDB collections to recommended subsets (e.g., 'h', 'c2', 'c5') using
subsetCollection to reduce noise.
- Set a random seed using
set.seed before plotting networks (plotMsigNetwork) or PPIs to ensure reproducible layouts.
- Combine multiple visualization panels (word clouds, networks, gene stats, and PPIs) using
patchwork (+ and plot_layout) for collective interpretation.
Common Pitfalls
- Unmatched gene identifiers: Ensure the gene IDs in your gene-level statistics vector match the ID type (e.g., Symbol vs Ensembl) used in the
GeneSetCollection.
- Extremely dense networks: If the overlap network is too crowded, increase the
thresh parameter in computeMsigOverlap to filter out weak overlaps.
Alternatives
enrichplot for alternative visualization of enrichment results.
clusterProfiler for functional profiling and gene set enrichment analysis.
Citations
- Bhuva DD (2026). vissE: Visualising Set Enrichment Analysis Results. R package.
References