groupFGSEA: Cluster gene set analysis results

In calebclass/NanoTube: An Easy Pipeline for NanoString nCounter Data Analysis

Cluster gene set analysis results

Description

Groups the pathway analysis results (using limmaToFGSEA or nsdiffToFGSEA) based on the enriched gene sets' leading edges. If the calculated distance metric is lower than the given threshold (i.e. the gene sets have highly overlapping leading edge genes), these gene sets will be joined to a single gene set "cluster." Or if 'ngroups' is specified, gene sets will be clustered by similarity into that number of groups.

Usage

groupFGSEA(
  gsea.res,
  l.edge,
  join.threshold = NULL,
  ngroups = NULL,
  dist.method = "binary",
  returns = c("signif", "all")
)

Arguments

gsea.res	Results from pathway analysis for a single comparison, using limmaToFSEA.
l.edge	Leading edge result from fgseaToLEdge.
join.threshold	The threshold distance to join gene sets. Gene sets with a distance below this value will be joined to a single "cluster."
ngroups	The desired number of gene set groups. Either 'join.threshold' or 'ngroups' must be specified, 'ngroups' takes priority if both are specified.
dist.method	Method for distance calculation (see options for dist()). We recommend the 'binary' (also known as Jaccard) distance.
returns	Either "signif" or "all". This argument defines whether only significantly enriched gene sets are included in the output table, or if the full results are included. Regardless of this selection, only significantly enriched gene sets are clustered.

Value

A data frame including the FGSEA results, plus two additional columns for the clustering results:

Cluster	The cluster that the gene set was assigned to. Gene sets in the same cluster have a distance below the join.threshold.
best	Whether the gene set is the most enriched (by p-value) in a given cluster.

Examples

data("ExamplePathways")
data("ExampleResults") # Results from runLimmaAnalysis

fgseaResults <- limmaToFGSEA(ExampleResults, gene.sets = ExamplePathways,
                             min.set = 5, rank.by = "t")

leadingEdge <- fgseaToLEdge(fgseaResults, cutoff.type = "padj", 
                            cutoff = 0.25)

# Group the results, and only returns those satisfying the cutoff specified 
# in leadingEdge()
groupedResults <- groupFGSEA(fgseaResults$Autoimmune.retinopathy, 
                             leadingEdge$Autoimmune.retinopathy,
                             join.threshold = 0.5,
                             returns = "signif")

calebclass/NanoTube documentation built on Nov. 21, 2023, 12:31 p.m.