Skip to contents

Extract normalised gene expression for plotting

extract_gene_expression.Rd

Extracts dense normalised (log1p) expression values for a set of genes, optionally with additional observation metadata columns.

Usage

extract_gene_expression(
  object,
  features,
  obs_cols = NULL,
  scale = FALSE,
  clip = NULL,
  modality = c("rna", "adt"),
  layer = c("norm", "magic")
)

Arguments

object

A single cell class.

features

Character vector. Gene IDs to extract.

obs_cols

Optional character vector. Column names from the obs table to include.

scale

Boolean. Whether to z-score the expression values.

clip

Optional numeric. If scale = TRUE, clip z-scores to [-clip, clip].

modality

String. One of c("rna", "adt"). ADT is only available for SingleCellsMultiModal.

layer

String. One of c("norm", "magic"). With "magic" the values come from the imputed layer run_magic_sc() wrote, which only holds the genes it was asked for. Imputation inflates gene-gene correlation, so this is for looking at things, not for measuring them. Note that extract_dot_plot_data() deliberately has no such argument: group means of imputed values are exactly the quantity MAGIC manufactures.

Value

A data.table with a cell_id column, one column per gene, and any requested obs columns.

Examples

# normalised expression of three genes with a cell annotation
sc <- demo_single_cells()
dt <- extract_gene_expression(
  sc,
  features = get_gene_names(sc)[1:3],
  obs_cols = "cell_grp"
)
head(dt, 3)
#>     cell_id  gene_01  gene_02  gene_03    cell_grp
#>      <char>    <num>    <num>    <num>      <char>
#> 1: cell_001 6.222656 6.761719 5.531250 cell_type_1
#> 2: cell_002 4.113281 4.511719 4.113281 cell_type_2
#> 3: cell_003 4.300781 4.300781 3.226562 cell_type_3

unlink(sc@dir_data, recursive = TRUE, force = TRUE)