Add Zarr to benchmarks

.Rbuildignore

@@ -18,3 +18,4 @@
 ^data$
 ^codecov\.yml$
 ^benchmarks$
+^\.venv$

.github/workflows/lint.yaml

@@ -19,9 +19,10 @@ jobs:
     steps:
       - uses: actions/checkout@v6
 
-      - uses: r-lib/actions/setup-r@v2
+      - name: Setup R and Bioconductor
+        uses: grimbough/bioc-actions/setup-bioc@v1
         with:
-          use-public-rspm: true
+          bioc-version: devel
 
       - name: Install air
         run: curl -LsSf https://github.com/posit-dev/air/releases/latest/download/air-installer.sh | sh

.github/workflows/pkgdown.yaml

@@ -36,9 +36,10 @@ jobs:
         with:
           python-version: "3.x"
 
-      - uses: r-lib/actions/setup-r@v2
+      - name: Setup R and Bioconductor
+        uses: grimbough/bioc-actions/setup-bioc@v1
         with:
-          use-public-rspm: true
+          bioc-version: devel
 
       - uses: r-lib/actions/setup-r-dependencies@v2
         with:

.github/workflows/pr-commands.yaml

@@ -24,9 +24,10 @@ jobs:
         with:
           repo-token: ${{ secrets.GITHUB_TOKEN }}
 
-      - uses: r-lib/actions/setup-r@v2
+      - name: Setup R and Bioconductor
+        uses: grimbough/bioc-actions/setup-bioc@v1
         with:
-          use-public-rspm: true
+          bioc-version: devel
 
       - uses: r-lib/actions/setup-r-dependencies@v2
         with:

.github/workflows/test-coverage.yaml

@@ -26,10 +26,10 @@ jobs:
           sudo apt-get update
           sudo apt-get -y install hdf5-tools libsz2 libaec-dev
 
-      - name: Setup R
-        uses: r-lib/actions/setup-r@v2
+      - name: Setup R and Bioconductor
+        uses: grimbough/bioc-actions/setup-bioc@v1
         with:
-          use-public-rspm: true
+          bioc-version: devel
 
       - name: Install R dependencies
         uses: r-lib/actions/setup-r-dependencies@v2

.gitignore

@@ -63,4 +63,4 @@ benchmarks/results_*.txt
 vignettes/data/*.h5ad
 /doc/
 /Meta/
-/data/
+/data/

DESCRIPTION

@@ -49,6 +49,7 @@ Suggests:
     knitr,
     processx,
     rhdf5 (>= 2.52.1),
+    Rarr (>= 1.11.12),
     rmarkdown,
     S4Vectors,
     Seurat,

NAMESPACE

@@ -17,7 +17,9 @@ export(as_AnnData)
 export(generate_dataset)
 export(get_generator_types)
 export(read_h5ad)
+export(read_zarr)
 export(write_h5ad)
+export(write_zarr)
 importFrom(Matrix,as.matrix)
 importFrom(Matrix,sparseMatrix)
 importFrom(Matrix,t)
@@ -35,3 +37,4 @@ importFrom(reticulate,r_to_py)
 importFrom(rlang,`%||%`)
 importFrom(rlang,caller_env)
 importFrom(stats,setNames)
+importFrom(utils,tail)

R/AbstractAnnData.R

@@ -290,6 +290,37 @@ AbstractAnnData <- R6::R6Class(
       )
     },
     #' @description
+    #' Convert to a [`ZarrAnnData`]
+    #'
+    #' See [as_ZarrAnnData()] for more details on the conversion
+    #'
+    #' @param file See [as_ZarrAnnData()]
+    #' @param compression See [as_ZarrAnnData()]
+    #' @param mode See [as_ZarrAnnData()]
+    #'
+    #' @return A [`ZarrAnnData`] object
+    as_ZarrAnnData = function(
+      file,
+      compression = c(
+        "none",
+        "gzip",
+        "blosc",
+        "zstd",
+        "lzma",
+        "bz2",
+        "zlib",
+        "lz4"
+      ),
+      mode = c("w-", "r", "r+", "a", "w", "x")
+    ) {
+      as_ZarrAnnData(
+        adata = self,
+        file = file,
+        compression = compression,
+        mode = mode
+      )
+    },
+    #' @description
     #' Write the `AnnData` object to an H5AD file
     #'
     #' See [write_h5ad()] for details
@@ -313,6 +344,37 @@ AbstractAnnData <- R6::R6Class(
         chunk_size = chunk_size,
         mode = mode
       )
+    },
+    #' @description
+    #' Write the `AnnData` object to a Zarr file
+    #'
+    #' See [write_zarr()] for details
+    #'
+    #' @param path See [write_zarr()]
+    #' @param compression See [write_zarr()]
+    #' @param mode See [write_zarr()]
+    #'
+    #' @return `path` invisibly
+    write_zarr = function(
+      path,
+      compression = c(
+        "none",
+        "gzip",
+        "blosc",
+        "zstd",
+        "lzma",
+        "bz2",
+        "zlib",
+        "lz4"
+      ),
+      mode = c("w-", "r", "r+", "a", "w", "x")
+    ) {
+      write_zarr(
+        object = self,
+        path,
+        compression = compression,
+        mode = mode
+      )
     }
   ),
   private = list(

R/AnnData-usage.R

@@ -13,6 +13,7 @@
 #'
 #' - [InMemoryAnnData] stores data in memory
 #' - [HDF5AnnData] provides an interface to a H5AD file
+#' - [ZarrAnnData] provides an interface to a Zarr store
 #' - [ReticulateAnnData] wraps a Python `AnnData` object via \pkg{reticulate}
 #'
 #' See the class documentation for details.
@@ -89,6 +90,7 @@
 #'   \item{`as_Seurat()`}{Convert to [`SeuratObject::Seurat`], see [as_Seurat()]}
 #'   \item{`as_InMemoryAnnData()`}{Convert to [`InMemoryAnnData`], as [as_InMemoryAnnData()]}
 #'   \item{`as_HDF5AnnData()`}{Convert to [`HDF5AnnData`], see [as_HDF5AnnData()]}
+#'   \item{`as_ZarrAnnData()`}{Convert to [`ZarrAnnData`], see [as_ZarrAnnData()]}
 #'   \item{`as_ReticulateAnnData()`}{Convert to [`ReticulateAnnData`], see [as_ReticulateAnnData()]}
 #' }
 #'
@@ -102,6 +104,14 @@
 #'   }
 #' }
 #'
+#' \describe{
+#'   \item{
+#'     `write_zarr()`
+#'   }{
+#'     Write the `AnnData` object to a Zarr store, see [write_zarr()]
+#'   }
+#' }
+#'
 #' ## General methods:
 #'
 #' \describe{
@@ -113,6 +123,7 @@
 #' \describe{
 #'   \item{[AnnData()]}{Create an [InMemoryAnnData] object}
 #'   \item{[read_h5ad()]}{Read an `AnnData` from a H5AD file}
+#'   \item{[read_zarr()]}{Read an `AnnData` from a Zarr store}
 #'   \item{[as_AnnData()]}{Convert other objects to an `AnnData` object}
 #' }
 #'
@@ -122,6 +133,7 @@
 #'   inherit from
 #' @seealso [InMemoryAnnData] for the in-memory implementation of `AnnData`
 #' @seealso [HDF5AnnData] for the HDF5-backed implementation of `AnnData`
+#' @seealso [ZarrAnnData] for the Zarr-backed implementation of `AnnData`
 #' @seealso [ReticulateAnnData] for the reticulate-based implementation that wraps Python AnnData objects
 #'
 #' @name AnnData-usage

R/Rarr_utils.R

@@ -0,0 +1,105 @@
+# Zarr metadata files used to identify valid Zarr nodes (arrays or groups)
+ZARR_METADATA_FILES <- c(".zarray", ".zattrs", ".zgroup", "zarr.json")
+
+#' create_zarr_group
+#'
+#' Create a Zarr group
+#'
+#' @param store The location of the Zarr store
+#' @param name Name of the group
+#' @param version Zarr version
+#'
+#' @return `NULL`
+#'
+#' @noRd
+create_zarr_group <- function(store, name, version = "v2") {
+  # Split "a/b/c" into c("a", "b", "c")
+  split_name <- strsplit(name, split = "/", fixed = TRUE)[[1]]
+  if (length(split_name) > 1) {
+    # Build cumulative paths: c("a", "a/b", "a/b/c")
+    split_name <- vapply(
+      seq_along(split_name),
+      function(x) paste(split_name[seq_len(x)], collapse = "/"),
+      FUN.VALUE = character(1)
+    )
+    # Keep only the target and its immediate parent:
+    # split_name[1] = "a/b/c" (target), split_name[2] = "a/b" (parent)
+    split_name <- rev(tail(split_name, 2))
+    # Recursively ensure the parent group exists before creating the target
+    if (!dir.exists(file.path(store, split_name[2]))) {
+      create_zarr_group(store = store, name = split_name[2])
+    }
+  }
+  dir.create(file.path(store, split_name[1]), showWarnings = FALSE)
+  switch(
+    version,
+    v2 = {
+      write(
+        "{\"zarr_format\":2}",
+        file = file.path(store, split_name[1], ".zgroup")
+      )
+    },
+    v3 = {
+      cli_abort("Currently only zarr v2 is supported!")
+    },
+    cli_abort("Only zarr v2 is supported. Use version = 'v2'")
+  )
+}
+
+#' create_zarr
+#'
+#' Create Zarr store
+#'
+#' @param store The location of the Zarr store
+#' @param version Zarr version
+#'
+#' @return `NULL`
+#'
+#' @noRd
+create_zarr <- function(store, version = "v2") {
+  prefix <- basename(store)
+  dir <- gsub(paste0(prefix, "$"), "", store)
+  create_zarr_group(store = dir, name = prefix, version = version)
+}
+
+#' is_zarr_empty
+#'
+#' Check if a Zarr store is empty
+#'
+#' @param store The location of the Zarr store
+#'
+#' @return Returns `TRUE` if the Zarr store is empty
+#'
+#' @noRd
+is_zarr_empty <- function(store) {
+  files <- list.files(store, recursive = FALSE, full.names = FALSE)
+  all(files %in% ZARR_METADATA_FILES)
+}
+
+#' Zarr path exists
+#'
+#' Check that a path in Zarr exists
+#'
+#' @return Whether the `target_path` exists in `store`
+#' @noRd
+#'
+#' @param store Path to a Zarr store
+#' @param target_path The path within the store to test for
+zarr_path_exists <- function(store, target_path) {
+  zarr <- file.path(store, target_path)
+  if (!dir.exists(zarr)) {
+    FALSE
+  } else {
+    list_files <- list.files(
+      path = zarr,
+      full.names = FALSE,
+      recursive = FALSE,
+      all.files = TRUE
+    )
+    if (any(ZARR_METADATA_FILES %in% list_files)) {
+      TRUE
+    } else {
+      FALSE
+    }
+  }
+}