Skip to contents

Derive sparse module membership from a loading matrix

modules_from_loadings.Rd

Turns a gene x k loading matrix from a matrix factorisation (ICA, NMF, DGRDL) into a membership table by keeping the tails of each component's loading distribution.

Two thresholding rules, selected via membership_params:

  • "zscore" - standardisation per component, keeping abs(z) > cutoff. No distributional assumption beyond rough symmetry.

  • "fdr" - two-sided p-values against a Normal null fitted per component, Benjamini-Hochberg adjusted, keeping padj < fdr.

The standardisation itself is controlled by membership_params$scaling: "robust" centres and scales by the median and MAD, "standard" by the mean and standard deviation. The latter is stricter and keeps fewer genes on skewed loadings, which is common for NMF.

Usage

modules_from_loadings(loadings, membership_params = params_module_membership())

Arguments

loadings

Numeric matrix. gene x k, with row and column names.

membership_params

List. See params_module_membership().

Value

A data.table with columns gene, module_id, loading, sign and the per-component score (z or padj depending on the method). One row per surviving (gene, component) pair, ordered by component then by descending absolute loading.

Examples

# turn NMF gene loadings into module membership
syn <- generate_gene_module_data(n_samples = 24L, n_genes = 60L)
# NMF needs a non-negative matrix
mat <- syn$data - min(syn$data)
obj <- BulkCoExp(mat, syn$meta_data)
obj <- preprocess_bulk_coexp(
  obj, hvg = NULL, scaling = FALSE, .verbose = FALSE
)
obj <- nmf_bulk(obj, k = 4L, .verbose = FALSE)
head(modules_from_loadings(get_nmf_gene_loadings(obj)))
#>          gene module_id  loading   sign        z
#>        <char>    <char>    <num> <char>    <num>
#> 1: feature_44   comp_01 3.675330    pos 38.28581
#> 2: feature_40   comp_01 3.643421    pos 37.94433
#> 3: feature_42   comp_01 3.633641    pos 37.83967
#> 4: feature_31   comp_01 3.630990    pos 37.81130
#> 5: feature_34   comp_01 3.602253    pos 37.50377
#> 6: feature_45   comp_01 3.590235    pos 37.37517