Construct cell groups based on TF networks and deconvolution

Identifies and projects cell groups using module relationships derived from TF networks and deconvolution outputs. If a binary trait is specified, the function splits the data and constructs cell groups for both classes (supervised analysis).

Usage

construct_cell_groups(
  network,
  dt,
  batch = NULL,
  pval = 0.05,
  clustering.method = "ward.D2",
  n_perm = 999,
  dendrogram_file = NULL,
  return_dendrogram = FALSE
)

Arguments

network

A list containing TF networks for cell types.

dt

A list containing deconvolution subgroup structures.

batch

Optional vector indicating batch assignment for samples.

pval

Numeric. P-value threshold applied both to filter TF module-deconvolution feature correlations and as the significance cutoff for the CCA permutation test. Default: 0.05.

clustering.method

Clustering method for hierarchical clustering. Default: "ward.D2".

n_perm

Integer. Number of permutations for the CCA significance test per cell group. Higher values give more precise p-values but increase runtime. Default: 999.

dendrogram_file

Optional character. File path to save dendrogram plot output.

return_dendrogram

Logical. If TRUE, includes the dendrogram in the returned list. Default FALSE.

Value

A list of 3 elements:

scores

A data frame or matrix with the projected cell group scores (samples x groups).

composition

A named list where each element is a character vector of original cell types per group.

loadings

A list of numeric vectors indicating the loadings (feature contributions) for each group.