Skip to contents

Module-Trait Correlation

Source: R/ModuleTraitCorrelation.R
ModuleTraitCorrelation.Rd

Computes the correlation between WGCNA modules (eigengenes or scores) and biological/clinical traits (metadata columns) for single cells.

Usage

ModuleTraitCorrelation(
  seurat_obj,
  traits,
  group.by = NULL,
  features = "hMEs",
  cor_method = "pearson",
  subset_by = NULL,
  subset_groups = NULL,
  wgcna_name = NULL,
  ...
)

Arguments

seurat_obj

A Seurat object containing the hdWGCNA experiment.

traits

A character vector of column names in seurat_obj@meta.data to correlate with each module. Traits must be numeric, integer, or factor. Character vectors should be converted to factors before running this function.

group.by

A string containing the name of a column in seurat_obj@meta.data. If provided, correlations are computed separately for each group (e.g., per cell type). If NULL (default), all cells are correlated together (global).

features

Character; which feature to use to summarize each module?

  • "hMEs": Harmonized Module Eigengenes

  • "MEs": Standard Module Eigengenes.

  • "scores": Module Scores (Seurat AddModuleScore).

cor_method

Character; method used for correlation. Valid choices: "pearson", "spearman", "kendall".

subset_by

A string containing the name of a column to subset the data before correlation.

subset_groups

A character vector specifying which groups from subset_by to include in the analysis.

wgcna_name

The name of the hdWGCNA experiment in seurat_obj@misc. Default = NULL (uses active WGCNA).

...

Additional arguments passed to internal functions.

Value

A Seurat object with the module-trait correlation results added to seurat_obj@misc[[wgcna_name]]$mt_cor. The results include correlation matrices, p-values, and FDR-corrected p-values for all cells and for each group specified in group.by.

Details

This function calculates the correlation between module eigengenes and user-specified traits. If a trait is a factor, it is converted to numeric based on the order of its levels. The function computes p-values and False Discovery Rates (FDR) for each correlation.