Run CIBERSORT deconvolution

Estimate immune cell proportions from a bulk expression matrix using the external CIBERSORT package or the native scop C++ backend. sig_matrix = "LM22" downloads the LM22 signature matrix from mengxu98/datasets and caches it locally.

Usage

RunCIBERSORT(
  object = NULL,
  count_matrix = NULL,
  sig_matrix = "LM22",
  bulk_assay = "counts",
  perm = 100,
  QN = TRUE,
  absolute = FALSE,
  backend = c("r", "cpp"),
  n_threads = 1L,
  seed = 123L,
  verbose = TRUE,
  ...
)

Arguments

object

Optional SummarizedExperiment object or expression matrix. When a SummarizedExperiment is provided, results are stored in metadata(object)[["Deconvolution"]].

count_matrix

Optional expression matrix with genes in rows and samples in columns. Used when object is not provided as a matrix.

sig_matrix

Signature matrix, local file path, or "LM22".

bulk_assay

Assay name in object used as the bulk counts matrix.

perm

Number of CIBERSORT permutations.

QN

Whether CIBERSORT should use quantile normalization.

absolute

Passed to CIBERSORT when supported by the installed package. The native C++ backend currently returns relative fractions.

backend

CIBERSORT backend. "r" calls the external CIBERSORT package and "cpp" uses the native scop LIBSVM implementation.

n_threads

Number of threads used by the C++ backend.

seed

Random seed used by the C++ permutation backend.

verbose

Whether to print the message. Default is TRUE.

...

Additional parameters forwarded to the internal deconvolution backend.

Value

A deconvolution result bundle for matrix input, or the modified SummarizedExperiment object for SummarizedExperiment input.

Examples

data(islet_bulk)

if (FALSE) {
# Run CIBERSORT
islet_bulk <- RunCIBERSORT(
  object = islet_bulk,
  sig_matrix = "LM22",
  bulk_assay = "counts",
  perm = 100,
  QN = TRUE
)

# Immune abundance stacked bar plot
p1 <- ImmuneAbundancePlot(
  object = islet_bulk,
  plot_type = "bar",
  group.by = "condition"
)
p1

# Immune cell correlation heatmap
p2 <- ImmuneAbundancePlot(
  object = islet_bulk,
  plot_type = "cor"
)
p2

# Gene-immune correlation butterfly plot
p3 <- GeneImmuneCorPlot(
  object = islet_bulk,
  features = rownames(SummarizedExperiment::assay(islet_bulk, "counts"))[1:3]
)
p3
}