Perform an immune cell deconvolution on a dataset.
deconvolute( gene_expression, method = deconvolution_methods, indications = NULL, tumor = TRUE, arrays = FALSE, column = "gene_symbol", rmgenes = NULL, scale_mrna = TRUE, expected_cell_types = NULL, ... )
A gene expression matrix or a Biobase ExpressionSet.
Either: A numeric matrix or data.frame with HGNC gene symbols as rownames and sample identifiers as colnames.
Or: A Biobase ExpressionSet with HGNC symbols in an fData column (see
In both cases, data must be on non-log scale.
a string specifying the method.
Supported methods are
a character vector with one indication per sample for TIMER. Argument is ignored for all other methods.
use a signature matrix/procedure optimized for tumor samples, if supported by the method. Currently affects EPIC and quanTIseq.
Runs methods in a mode optimized for microarray data. Currently affects quanTIseq and CIBERSORT.
Only relevant if
gene_expression is an ExpressionSet. Defines in which column
of fData the HGNC symbol can be found.
a character vector of gene symbols. Exclude these genes from the analysis. Use this to exclude e.g. noisy genes.
logical. If FALSE, disable correction for mRNA content of different cell types. This is supported by methods that compute an absolute score (EPIC and quanTIseq)
Limit the analysis to the cell types given in this list. If the cell types present in the sample are known a priori, setting this can improve results for xCell (see https://github.com/grst/immunedeconv/issues/1).
arguments passed to the respective method
cell_type as first column and a column with the
calculated cell fractions for each sample.
# Not run: deconvolute(gene_expression_matrix, "xcell")