Compute associations between TF module scores and clinical metadata

This function tests for associations between transcription factor (TF) module scores and available clinical traits. It uses Pearson correlation for continuous (numeric) traits, and ANOVA for categorical traits. Results are visualized as a labeled heatmap and violin plots. All plots are saved in the Results/ directory.

Usage

# S3 method for class 'metadata.association'
compute(
  tfs.modules,
  coldata,
  pval = 0.05,
  corr_method = "p",
  file.name,
  width = 20,
  height = 8,
  ncol = 5,
  y_min = 0,
  y_max = 0.5,
  plot_grid = F,
  width_grid = 18,
  height_grid = 10
)

Arguments

tfs.modules

A numeric matrix or data frame of TF module scores across samples. Typically the output from compute.WTCNA(). Rows represent samples, columns represent TF modules.

coldata

A data frame containing clinical traits (both categorical and numerical) for the same samples. Row names should match those of tfs.modules.

pval

A numeric threshold (default = 0.05) to determine statistical significance. Only associations with p-values below this threshold are considered significant in the heatmap.

corr_method

Character string specifying the correlation method to use for continuous variables. Options are "p" for pearson and "s" for spearman. Default is "p".

file.name

Character. Base file name for saving PDF plots of results.

width

A numeric value indicating the width (in inches) of the output heatmap plot (default = 20).

height

A numeric value indicating the height (in inches) of the output heatmap plot (default = 8).

ncol

Integer. Number of columns in the grid of association boxplots.

y_min

Numeric. Lower y-axis limit for grid boxplots.

y_max

Numeric. Upper y-axis limit for grid boxplots.

plot_grid

Logical; if TRUE, generates a grid of boxplot summaries.

width_grid

Numeric width of the grid plot output.

height_grid

Numeric height of the grid plot output.

Value

This function saves the following to the Results/ directory:

• A labeled heatmap showing Pearson correlations and ANOVA test p-values.

• Individual violin plots for significant categorical trait associations.

The function does not return an object to the R environment.

Examples

data("network.tuto")
data("traitdata.tuto")

compute.metadata.association(
  tfs.modules = network.tuto[[1]],
  coldata = traitdata.tuto,
  pval = 0.05,
  file.name = 'Tutorial',
  width = 15,
  height = 10
)