Kruskal-Wallis test for multi-group comparisons

Performs a Kruskal-Wallis test to compare scores across multiple trait levels. Significant results are visualized as annotated boxplots with Dunn post-hoc tests.

Usage

scores.kruskal.test(scores, coldata, trait, pval = 0.05)

Arguments

scores

A list, NMF output from compute.latent_factors(), or a score matrix. When a list, the first element must be a samples x features score matrix.

coldata

A data frame containing sample annotations, including the grouping trait.

trait

Character. Name of the column in coldata used as the grouping variable.

pval

Numeric. P-value threshold for significance (default = 0.05).

Value

A list containing only significant features after Kruskal-Wallis test. Returns NULL if no significant features are found.