Write annotations and uns to csv files — write

Writes obs, var, obsm, varm and uns (and X when skip_data is FALSE) to csv files in a directory, as Python's write_csvs() does. obs.csv and var.csv have a header and their names as the first column, obsm.csv and varm.csv have one column per column of each matrix (named by the key and the 1-based column number) and no names, and X.csv and uns/<key>.csv have neither header nor names. Numbers are written as pandas does, so reading the files with pandas gives the same values whichever language wrote them.

Usage

write_csvs(adata, dirname, skip_data = TRUE, sep = ",")

Arguments

adata: An AnnData() object.
dirname: Directory to write to, created if needed. A .csv suffix is removed.
skip_data: Whether to leave out X.
sep: The field separator.

Value

NULL, invisibly.

Details

Sparse uns values are omitted with one anndata_write_warning, and so is any uns value with more than 2 dimensions. A named list in uns (a dict) is written as one cell holding Python's repr of the dict, which supports strings, scalars and nested dicts. A View is written as its subset and stays a View.

Examples

adata <- AnnData(X = matrix(c(1, 2, 3, 4), 2, 2))
dir <- tempfile("csvs")
write_csvs(adata, dir, skip_data = FALSE)
list.files(dir, recursive = TRUE)
#> [1] "X.csv"    "obs.csv"  "obsm.csv" "var.csv"  "varm.csv"
unlink(dir, recursive = TRUE)