Skip to contents

Analyze Fuzzy C-means Clustering Results with ORA

run_cluster_ORA.Rd

Over-representation analysis (ORA) applied to results of fuzzy c-means (FCM) clustering.

Usage

run_cluster_ORA(
  FCM,
  selected_omes = c("transcript-rna-seq", "prot-pr", "prot-ph", "metab"),
  selected_tissues = "all",
  database = names(MotrpacHumanPreSuspensionAnalysis::MOLECULAR_SIGNATURES),
  path_to_gmt = NULL,
  min_size = 5L,
  overlap_cutoff = 0.7,
  min_prob = 0.3
)

Arguments

FCM

fuzzy c-means (FCM) clustering results. Output of run_cmeans. This should be a named list of objects of class fclust.

selected_omes

character; one or more character strings selected from the following options: "transcript-rna-seq", "prot-pr", "prot-ph", and "metab" (all metabolomics platforms). Passed to load_differential_analysis.

selected_tissues

character; passed to load_differential_analysis. One or more of the following: "all", "muscle", "adipose", or "blood".

database

character; one or more names specifying the database(s) to test. Options are (case insensitive) "BIOCARTA", "KEGG_MEDICUS", "PID", "REACTOME", "WP" (WikiPathways database), "GOBP", "GOCC", "GOMF", "MITOCARTA" (MitoCarta3.0 database), "PSP" (PhosphoSitePlus kinases; only valid when selected_omes contains "prot-ph"), or "REFMET" (RefMet chemical subclasses; only valid when selected_omes contains "metab"). See MOLECULAR_SIGNATURES for details.

path_to_gmt

character; (optional) path to one or more GMT files. Passed to TMSig::readGMT. If provided, database is ignored.

min_size

integer; the minimum set size for testing.

overlap_cutoff

numeric; the minimum proportion of genes in each set that must appear in a given dataset. Used to pre-filter sets. Does not affect "metab" or "prot-ph" results. This will always be 0.1 for "prot-ol" results.

min_prob

numeric; the minimum cluster membership probability required for features to belong to a cluster. All features will be used for the background for ORA, regardless of membership probability.

Value

An object of class data.frame with the following columns:

tissue

factor; the tissue.

assay

factor; the omics assay.

cluster

factor; the cluster number. The clusters have the same meaning across omics assays measured in the same tissue, but they have no such relationship across tissues.

collection

factor; the broad molecular signature collection. See SET_TO_ID for details.

database

factor; the molecular signature database. See SET_TO_ID for details.

set_id

character; a unique ID for the molecular signature. See SET_TO_ID for details.

set

character; the molecular signature being tested. For global proteomics and transcriptomics, these are gene sets. For phosphoproteomics, these are kinase sets.

set_short

character; a shortened version of set. See SET_TO_ID for details.

set_size

integer; the number of molecules in the set that were present in the DA results for that specific tissue/assay combination.

set_size_DB

integer; the number of molecules in the set, as defined in the GMT file.

size_ratio

numeric; the ratio of set_size to set_size_DB, rounded to the nearest thousandth. A measure of confidence that the gene set being tested is correctly described by the entry in the set column. While smaller values do not necessarily indicate that the results are unreliable, terms from the gene set databases should be treated with caution.

set_size_in_cluster

integer; the number of elements in the set that localized to the cluster.

cluster_size

integer; number of features in each cluster with membership probabilities of at least min_prob.

background_size

integer; the number of elements in the background. Includes all features that appeared in clustering results.

p_value

numeric; the upper-tail p-value.

adj_p_value

numeric; the BH-adjusted p-value. P-values are adjusted within each combination of tissue, assay, collection, and cluster.

See also

run_cmeans, run_cluster_cameraPR, MOLECULAR_SIGNATURES, SET_TO_ID

Author

Tyler Sagendorf

Examples

if (FALSE) { # \dontrun{
  FCM <- run_cmeans()

  # Run ORA with all available molecular signatures
  cluster_res <- run_cluster_ORA(FCM = FCM)
  head(cluster_res)
} # }