Deconvolve bulk counts with BayesPrism — deconv

Both bulk and single-cell reference inputs must be raw integer counts. The MAE assay is unambiguously gene-by-sample and is converted to BayesPrism's sample-by-gene layout. reference_orientation explicitly declares the supplied reference layout; no orientation is inferred from dimensions.

Usage

deconv_bayesprism(
  x,
  experiment,
  reference,
  cell_type_labels = NULL,
  assay = "counts",
  reference_orientation = c("genes_by_cells", "cells_by_genes"),
  cell_state_labels = NULL,
  key = NULL,
  input_type = "count.matrix",
  cores = 1,
  update_gibbs = TRUE,
  gibbs_control = list(),
  opt_control = list(),
  ...,
  reference_assay = NULL
)

Arguments

x: A MultiAssayExperiment object.
experiment: Optional experiment name.
reference: Raw reference count matrix, or a SingleCellExperiment created by deconv_reference().
cell_type_labels: Cell-type label for each reference cell. For a SingleCellExperiment, NULL uses its cell_type column and a single string names another colData column.
assay: Assay name or one-based assay index.
reference_orientation: Layout of reference. The default preserves the package's documented gene-by-cell input, while cells_by_genes accepts BayesPrism's native layout directly.
cell_state_labels: Optional cell-state label for each reference cell. For a SingleCellExperiment, NULL uses cell_state when that column is present and a single string names another colData column.
key: Optional malignant/tumour key accepted by BayesPrism.
input_type: BayesPrism reference input type.
cores: Number of worker cores passed to run.prism().
update_gibbs: Run the final Gibbs update.
gibbs_control, opt_control: Named control lists passed to BayesPrism::run.prism().
...: Additional arguments passed to BayesPrism::new.prism().
reference_assay: Raw-count assay used when reference is a SingleCellExperiment. NULL selects counts, or the sole assay when no counts assay is present.

Value

The native object returned by BayesPrism::run.prism().