Do not copy items is mapping is empty vector `c()`

[LouiseDck]

library(anndataR)

anndata <- generate_dataset(
  n_obs = 10,
  n_vars = 20,
  format = "AnnData",
  uns_types = c()
)

sce <- to_SingleCellExperiment(anndata)
#> Registered S3 method overwritten by 'future':
#>   method               from      
#>   all.equal.connection parallelly
ad2 <- from_SingleCellExperiment(
  sce,
  x_mapping = "numeric_matrix",
  obs_mapping = c(),
  var_mapping = c(),
  obsm_mapping = list(),
  obsp_mapping = list(),
  varm_mapping = list(),
  varp_mapping = list(),
  uns_mapping = list(),
)
ad2
#> AnnData object with n_obs × n_vars = 10 × 20
#>     obs: 'character', 'integer', 'factor', 'factor_ordered', 'logical', 'numeric', 'character_with_nas', 'integer_with_nas', 'factor_with_nas', 'factor_ordered_with_nas', 'logical_with_nas', 'numeric_with_nas'
#>     var: 'character', 'integer', 'factor', 'factor_ordered', 'logical', 'numeric', 'character_with_nas', 'integer_with_nas', 'factor_with_nas', 'factor_ordered_with_nas', 'logical_with_nas', 'numeric_with_nas'
#>     layers: 'counts', 'numeric_dense', 'numeric_csparse', 'numeric_rsparse', 'numeric_matrix_with_nas', 'numeric_dense_with_nas', 'numeric_csparse_with_nas', 'numeric_rsparse_with_nas', 'integer_matrix', 'integer_dense', 'integer_csparse', 'integer_rsparse', 'integer_matrix_with_nas', 'integer_dense_with_nas', 'integer_csparse_with_nas', 'integer_rsparse_with_nas'

[LouiseDck]

We should update the documentation (probably after merging #253 )

[rcannood]

Make sure we:

• Switch to using either lists or vectors to work the same way, or we switch to using either one of the other

Options for notations:

Going back to lists

ad2 <- from_SingleCellExperiment(
  sce,
  x_mapping = "numeric_matrix",
  obs_mapping = list("foo", "bar"),
  var_mapping = list(),
  obsm_mapping = list(),
  obsp_mapping = list(),
  varm_mapping = list(),
  varp_mapping = list(),
  uns_mapping = list()
)

Update documentation

ad2 <- from_SingleCellExperiment(
  sce,
  x_mapping = "numeric_matrix",
  obs_mapping = c("foo", "bar"),
  var_mapping = character(0),
  obsm_mapping = character(0),
  obsp_mapping = character(0),
  varm_mapping = character(0),
  varp_mapping = character(0),
  uns_mapping = character(0)
)

TRUE/FALSE

ad2 <- from_SingleCellExperiment(
  sce,
  x_mapping = "numeric_matrix",
  obs_mapping = c("foo", "bar"),
  var_mapping = TRUE,
  obsm_mapping = TRUE,
  obsp_mapping = FALSE,
  varm_mapping = FALSE,
  varp_mapping = FALSE,
  uns_mapping = FALSE
)

Allow specifying the default

ad2 <- from_SingleCellExperiment(
  sce,
  x_mapping = "numeric_matrix",
  obs_mapping = c("foo", "bar"),
  var_mapping = character(0),
  obsm_mapping = character(0),
  obsp_mapping = character(0),
  varm_mapping = character(0),
  varp_mapping = character(0),
  uns_mapping = character(0),
  copy_all_by_default = TRUE # or whatever interface you want this to have
)

[rcannood]

Or support something like tidyselect https://dplyr.tidyverse.org/reference/select.html

[rcannood]

Let's go with TRUE / FALSE!

• TRUE: Copy everything
• FALSE: Copy nothing
• Empty vectors / empty lists / NULL means the same as FALSE
• Non-empty character vector: Copy just these items
• Named non-empty character vectors: Copy just these items and rename them

New default would be TRUE