run_gene_prop_celltype: Generate cell type specific gene proportion expression

In SydneyBioX/scFeatures: scFeatures: Multi-view representations of single-cell and spatial data for disease outcome prediction

Generate cell type specific gene proportion expression

Description

This function computes the proportion of expression of a set of genes for each cell type in the input data. The input data can be of three types: 'scrna', 'spatial_p' or 'spatial_t'. If the genes parameter is not provided by the user, the top variable genes will be selected based on the num_top_gene parameter (defaults to 100).

Usage

run_gene_prop_celltype(
  data,
  type = "scrna",
  genes = NULL,
  num_top_gene = NULL,
  ncores = 1
)

Arguments

data	A list object containing data matrix and celltype and sample vector.
type	The type of dataset, either "scrna", "spatial_t", or "spatial_p".
genes	Optional dataframe with 2 columns: 'marker' and 'celltype'. The 'marker' column should contain the genes of interest (e.g. 'S100A11', 'CCL4'), and the 'celltype' column should contain the celltype that the gene expression is to be computed from (e.g. 'CD8', 'B cells'). If not provided, the top variable genes will be used based on the num_top_gene parameter.
num_top_gene	Number of top genes to use when genes is not provided. Defaults to 100.
ncores	Number of cores for parallel processing.

Value

a dataframe of samples x features. The features are in the form of gene 1 celltype a, gene 2 celltype b ... etc, with the number representing proportion of gene expression of the given gene across the cells of the the given celltype.

Examples

utils::data("example_scrnaseq" , package = "scFeatures")
data <- example_scrnaseq[, 1:20]
celltype <- data$celltype
sample <- data$sample
data <- data@assays$RNA@data

alldata <- scFeatures:::formatData(data = data, celltype = celltype, sample = sample )

feature_gene_prop_celltype <- run_gene_prop_celltype(
    alldata,
    type = "scrna", num_top_gene = 100, ncores = 1
 )

SydneyBioX/scFeatures documentation built on Nov. 30, 2024, 6:40 p.m.