Skip to contents

Differential co-expression analysis with CS-CORE

diff_CSCORE.Rd

Test for differential co-expression between two groups using a permutation test. CS-CORE is run on each group separately to obtain co-expression estimates. Group labels are then permuted n_permu times at the biological sample level to build a null distribution of differences in co-expression estimates. Two-sided permutation p-values are returned for each gene pair.

Usage

diff_CSCORE(
  object,
  group_label,
  sample_label,
  genes,
  group_levels = NULL,
  n_permu = 100,
  seed = NULL,
  n_cores = 1L,
  verbose = FALSE,
  ...
)

Arguments

object

A Seurat object containing single-cell RNA-seq data, subsetted to a single cell type.

group_label

A character string giving the name of the object@meta.data column that contains the group labels (e.g. "Status").

sample_label

A character string giving the name of the object@meta.data column that identifies biological samples / donors (e.g. "Donor"). Permutations are performed at the sample level to preserve within-sample cell correlations.

genes

A character vector of gene names (length \(p\)) for which co-expression will be estimated.

group_levels

Optional. A length-2 character vector specifying the two group labels and their order, e.g. c("Healthy", "COVID"). The observed difference is computed as group 2 minus group 1. Defaults to sort(unique(groups)).

n_permu

Integer; number of permutations. Defaults to 100.

seed

Optional integer seed passed to set.seed() before selecting permutation combinations. Defaults to NULL.

n_cores

Integer; number of cores for parallelizing the permutation loop via parallel::mclapply. Defaults to 1L (sequential). Values greater than 1 are not supported on Windows.

verbose

Logical; if FALSE (default), suppresses all output from the underlying CSCORE calls and prints only a progress bar for the permutation loop. If TRUE, prints detailed [INFO] messages from each step.

...

Additional arguments passed to CSCORE.

Value

A list with four elements:

est_group1

A \(p \times p\) matrix of co-expression estimates for group 1.

est_group2

A \(p \times p\) matrix of co-expression estimates for group 2.

obs_diff

A \(p \times p\) matrix of observed differences in co-expression (est_group2 - est_group1).

p_value

A \(p \times p\) matrix of two-sided permutation p-values. Diagonal entries are NA.

See also

Differential co-expression tutorial

Examples

# See a full example at:
# https://changsubiostats.github.io/CS-CORE/articles/differential_coexpression.html