Decompose bulk RNA-seq samples using a single cell reference

CIBERSORT is an algorithm (Newman et al. Nature Methods 12:453-457) for estimating the cell type composition of a bulk sample, given a gene expression profile of the sample and a known gene expression profile for each cell type potentially contributing to the sample. Here, we wrap runCIBERSORT function from RNAMagnet R package to decompose bulk RNA-seq samples. To install RNAMagnet R package, see https://github.com/veltenlab/rnamagnet. Note that MAGIC is an important requirement of RNAMagnet package, and needs to be installed as a python package, see https://github.com/KrishnaswamyLab/MAGIC/tree/master/Rmagic#installation.

Usage

decomposeBulk(
  SeuratObj,
  exprs,
  meta,
  mc.cores = 1,
  slot = "counts",
  assay = "SCT"
)

Arguments

SeuratObj

A Seurat object.

exprs

A data frame or matrix of raw read counts of bulk RNA-seq samples. Column names correspond to sample names, row names to genes.

meta

A data frame of 2 columns with Sample and Class, correlate sample names with sample class.

mc.cores

Number of cores used

slot

Which slot to use to calculate mean expression per cell type

assay

Which assays to use to calculate mean expression per cell type

Value

A data frame in long format, indicating the fraction of cell types in each bulk RNA-seq sample.