gs_fuzzyclustering: Compute fuzzy clusters of gene sets

In federicomarini/GeneTonic: Enjoy Analyzing And Integrating The Results From Differential Expression Analysis And Functional Enrichment Analysis

Compute fuzzy clusters of gene sets

Description

Compute fuzzy clusters of different gene sets, aiming to identify grouped categories that can better represent the distinct biological themes in the enrichment results

Usage

gs_fuzzyclustering(
  res_enrich,
  gtl = NULL,
  n_gs = nrow(res_enrich),
  gs_ids = NULL,
  similarity_matrix = NULL,
  similarity_threshold = 0.35,
  fuzzy_seeding_initial_neighbors = 3,
  fuzzy_multilinkage_rule = 0.5
)

Arguments

res_enrich	A data.frame object, storing the result of the functional enrichment analysis. See more in the main function, GeneTonic(), to check the formatting requirements (a minimal set of columns should be present).
gtl	A GeneTonic-list object, containing in its slots the arguments specified above: dds, res_de, res_enrich, and annotation_obj - the names of the list must be specified following the content they are expecting
n_gs	Integer value, corresponding to the maximal number of gene sets to be displayed
gs_ids	Character vector, containing a subset of gs_id as they are available in res_enrich. Lists the gene sets to be displayed.
similarity_matrix	A similarity matrix between gene sets. Can be e.g. computed with create_kappa_matrix() or create_jaccard_matrix() or a similar function, returning a symmetric matrix with numeric values (max = 1). If not provided, this will be computed on the fly with create_kappa_matrix()
similarity_threshold	A numeric value for the similarity matrix, used to determine the initial seeds as in the implementation of DAVID. Higher values will lead to more genesets being initially unclustered, leading to a functional classification result with fewer groups and fewer geneset members. Defaults to 0.35, recommended to not go below 0.3 (see DAVID help pages)
fuzzy_seeding_initial_neighbors	Integer value, corresponding to the minimum geneset number in a seeding group. Lower values will lead to the inclusion of more genesets in the functional groups, and may generate a lot of small size groups. Defaults to 3
fuzzy_multilinkage_rule	Numeric value, comprised between 0 and 1. This parameter will determine how the seeding groups merge with each other, by specifying the percentage of shared genesets required to merge the two subsets into one group. Higher values will give sharper separation between the groups of genesets. Defaults to 0.5 (50%)

Value

A data frame, shaped in a similar way as the originally provided res_enrich object, containing two extra columns: gs_fuzzycluster, to specify the identifier of the fuzzy cluster of genesets, and gs_cluster_status, which can specify whether the geneset is the "Representative" for that cluster or a simple "Member". Notably, the number of rows in the returned object can be higher than the original number of rows in res_enrich.

References

See https://david.ncifcrf.gov/helps/functional_classification.html#clustering for details on the original implementation

Examples

data(res_enrich_macrophage, package = "GeneTonic")
res_enrich <- shake_topGOtableResult(topgoDE_macrophage_IFNg_vs_naive)
# taking a smaller subset
res_enrich_subset <- res_enrich[1:100, ]

fuzzy_subset <- gs_fuzzyclustering(
  res_enrich = res_enrich_subset,
  n_gs = nrow(res_enrich_subset),
  gs_ids = NULL,
  similarity_matrix = NULL,
  similarity_threshold = 0.35,
  fuzzy_seeding_initial_neighbors = 3,
  fuzzy_multilinkage_rule = 0.5
)

# show all genesets members of the first cluster
fuzzy_subset[fuzzy_subset$gs_fuzzycluster == "1", ]

# list only the representative clusters
head(fuzzy_subset[fuzzy_subset$gs_cluster_status == "Representative", ], 10)

federicomarini/GeneTonic documentation built on March 27, 2024, 4:19 p.m.