Summarise cell metadata for each metacell. Metadata fields can be either numeric and then the summary function func is applied for the values of each field, or categorical metadata fields which are expanded to multiple
metadata columns with the fraction of cells (in each metacell) for every category. Variables that are either character,
factor or are explicitly set at categorical are treated as categorical.
cell_metadata_to_metacell converts cell metadata to metacell metadata from data frames. cell_metadata_to_metacell_from_h5ad extracts metadata fields and cell_to_metacell from cells h5ad file and
then runs cell_metadata_to_metacell. cell_metadata_to_metacell_from_metacell1 extracts metadata fields and cell_to_metacell from metacell1 scdb and
then runs cell_metadata_to_metacell.
Usage
cell_metadata_to_metacell(
cell_metadata,
cell_to_metacell,
func = mean,
categorical = c()
)
cell_metadata_to_metacell_from_metacell1(
scdb,
matrix,
mc,
metadata_fields,
func = mean,
categorical = c()
)
cell_metadata_to_metacell_from_h5ad(
anndata_file,
metadata_fields,
func = mean,
categorical = c(),
rm_outliers = TRUE
)Arguments
- cell_metadata
data frame with a column named "cell_id" with the cell id and other metadata columns, or a name of a delimited file which contains such data frame.
- cell_to_metacell
data frame with a column named "cell_id" with cell id and another column named "metacell" with the metacell the cell is part of, or a name of a delimited file which contains such data frame.
- func
summary function for the cell metadata for non categorical metadata columns (e.g. mean, median, sum)
- categorical
a vector with names of categorical variables. The returned data frame would have a column for each category where the values are the fraction of cells with the category in each metacell.
- scdb, matrix, mc
scdb, matrix and mc objects from metacell1. See
import_dataset_metacell1for more information.- metadata_fields
names of fields in the anndata
object$obswhich contains metadata for each cell.- anndata_file
path to
h5adfile which contains the output of metacell2 pipeline (metacells python package).- rm_outliers
do not calculate statistics for cells that are marked as outliers (
outiler=TRUEinobject$obs) (only relevant when runningcell_metadata_to_metacell_from_h5ad)
Value
A data frame with a column named "metacell" and
the metadata columns from cell_metadata summarized for each metacell using
func for non-categorical variables, and a column for each category of the categorical metadata variables
incell_metadata, where the values are the fraction of cells with the category in each metacell.
Examples
set.seed(60427)
n_cells <- 5e6
cell_metadata <- tibble::tibble(
cell_id = 1:n_cells,
md1 = sample(1:5, size = n_cells, replace = TRUE),
md2 = rnorm(n = n_cells),
md_categorical1 = sample(paste0("batch", 1:5), size = n_cells, replace = TRUE),
md_categorical2 = sample(1:5, size = n_cells, replace = TRUE)
)
cell_to_metacell <- tibble::tibble(
cell_id = 1:n_cells,
metacell = sample(0:1535, size = n_cells, replace = TRUE)
)
metadata <- cell_metadata_to_metacell(
cell_metadata[, 1:3],
cell_to_metacell
)
#> ℹ Numerical variables: md1, md2
head(metadata)
#> # A tibble: 6 x 3
#> metacell md1 md2
#> 1 0 2.979448 -0.021942785
#> 2 1 2.958991 0.015177899
#> 3 2 2.990102 0.005634195
#> 4 3 3.012686 -0.026841161
#> 5 4 3.005740 0.011632598
#> 6 5 2.960923 0.008044519
metadata1 <- cell_metadata_to_metacell(
cell_metadata[, 1:3], cell_to_metacell,
func = function(x) x * 2
)
#> ℹ Numerical variables: md1, md2
#> Warning: Returning more (or less) than 1 row per `summarise()` group was deprecated in dplyr 1.1.0.
#> ℹ Please use `reframe()` instead.
#> ℹ When switching from `summarise()` to `reframe()`, remember that `reframe()` always returns an
#> ungrouped data frame and adjust accordingly.
#> ℹ The deprecated feature was likely used in the dplyr package.
#> Please report the issue at <https://github.com/tidyverse/dplyr/issues>.
head(metadata1)
#> # A tibble: 6 x 3
#> metacell md1 md2
#> 1 0 6 -0.8932316
#> 2 0 2 -2.1180205
#> 3 0 10 2.2575215
#> 4 0 4 -0.7134182
#> 5 0 2 1.6701686
#> 6 0 6 1.4134616
metadata3 <- cell_metadata_to_metacell(
cell_metadata,
cell_to_metacell,
categorical = c("md_categorical1", "md_categorical2")
)
#> ℹ Categorical variables: md_categorical1, md_categorical2
#> ℹ Numerical variables: md1, md2
head(metadata3)
#> # A tibble: 6 x 13
#> metacell md_categorical1: batch1 md_categorical1: batch2
#> 1 0 0.1953988 0.2039877
#> 2 1 0.1876972 0.2015773
#> 3 2 0.2206258 0.1985951
#> 4 3 0.1921979 0.2029813
#> 5 4 0.1878189 0.2053571
#> 6 5 0.2058462 0.1932308
#> md_categorical1: batch3 md_categorical1: batch4 md_categorical1: batch5
#> 1 0.2046012 0.2058282 0.1901840
#> 2 0.2100946 0.2037855 0.1968454
#> 3 0.1928480 0.1973180 0.1906130
#> 4 0.2131304 0.1925151 0.1991754
#> 5 0.2050383 0.2005740 0.2012117
#> 6 0.2033846 0.1996923 0.1978462
#> md_categorical2: 1 md_categorical2: 2 md_categorical2: 3 md_categorical2: 4
#> 1 0.1935583 0.2141104 0.1966258 0.1889571
#> 2 0.2056782 0.2012618 0.1914826 0.2028391
#> 3 0.2043423 0.1979566 0.1985951 0.2097701
#> 4 0.2017127 0.2026641 0.1928322 0.2102759
#> 5 0.2117347 0.2037628 0.1903699 0.1951531
#> 6 0.2030769 0.1920000 0.2046154 0.2033846
#> md_categorical2: 5 md1 md2
#> 1 0.2067485 2.979448 -0.021942785
#> 2 0.1987382 2.958991 0.015177899
#> 3 0.1893359 2.990102 0.005634195
#> 4 0.1925151 3.012686 -0.026841161
#> 5 0.1989796 3.005740 0.011632598
#> 6 0.1969231 2.960923 0.008044519
if (FALSE) {
cell_metadata_to_metacell_from_h5ad("cells.h5ad", c("pile", "age", "batch"), categorical = "batch")
}