Skip to contents

Run bulk or pseudobulk deconvolution

Source: R/RunDeconvolution.R
RunDeconvolution.Rd

Estimate cell-type proportions from a bulk-like expression matrix stored in a SummarizedExperiment object, using a Seurat reference.

Usage

RunDeconvolution(object, ...)

# S3 method for class 'SummarizedExperiment'
RunDeconvolution(
  object,
  reference = NULL,
  method = c("MuSiC", "BisqueRNA", "BayesPrism", "CIBERSORT"),
  group.by = NULL,
  sample.by = NULL,
  cellstate.by = NULL,
  bulk_assay = "counts",
  ref_assay = NULL,
  ref_layer = "counts",
  backend = c("cpp", "r"),
  verbose = TRUE,
  ...
)

Arguments

object

A SummarizedExperiment object containing bulk-like counts.

...

Additional parameters forwarded to the internal deconvolution backend.

reference

A Seurat reference object used to build cell-type profiles. Not required for "CIBERSORT".

method

Deconvolution method. One of "MuSiC", "BisqueRNA", "BayesPrism", or "CIBERSORT".

group.by

Metadata column in reference defining reference cell types.

sample.by

Metadata column in reference defining biological sample / donor IDs. Used by the r backends of MuSiC and BisqueRNA. If NULL, SCOP will try to infer a suitable column automatically.

cellstate.by

Metadata column in reference defining cell states for the r backend of BayesPrism. If NULL, group.by is reused.

bulk_assay

Assay name in object used as the bulk counts matrix.

ref_assay

Assay name in reference used for the reference profiles.

ref_layer

Layer name in reference used for reference counts.

backend

Deconvolution engine backend. "r" uses the original method package implementation. "cpp" is reserved for native SCOP implementations when available.

verbose

Whether to print the message. Default is TRUE.

Value

A SummarizedExperiment object with results stored in S4Vectors::metadata(object)[["Deconvolution"]].

See also

DeconvolutionPlot

Examples

data(islet_bulk)
islet_bulk <- RunDeconvolution(
  islet_bulk,
  method = "CIBERSORT",
  backend = "cpp",
  perm = 0
)
#> ℹ [2026-07-02 09:36:47] Use 400 shared genes for CIBERSORT
DeconvolutionPlot(islet_bulk, plot_type = "bar")


DeconvolutionPlot(
  islet_bulk,
  plot_type = "heatmap",
  sample_annotation = "condition",
  sample_split = "condition"
)


DeconvolutionPlot(islet_bulk, plot_type = "box")