Cleanup vince

DESCRIPTION

@@ -1,7 +1,7 @@
 Package: SpatialData.data
 Title: SpatialData datasets
-Depends: R (>= 4.4), SpatialData
-Version: 0.99.2
+Depends: R (>= 4.4)
+Version: 0.99.3
 Description: Makes 'scverse' data examples available through the NSF OSN, 
     and accessible from within R, using 'BiocFileCache'. Furthermore, 
     provides an interface to Python's 'spatialdata-io' for reading and
@@ -18,15 +18,16 @@ Authors@R: c(
 Imports:
     basilisk,
     BiocFileCache,
-    reticulate
+    reticulate,
+    SpatialData
 Suggests:
     BiocStyle,
     knitr,
     magick,
-    paws,
     Rgraphviz,
     testthat,
-    DT
+    DT,
+    paws
 Remotes:
     HelenaLC/SpatialData
 biocViews: 

NAMESPACE

@@ -11,17 +11,11 @@ export(LungAdenocarcinomaMCMICRO)
 export(MouseBrainMERFISH)
 export(MouseIntestineVisHD)
 export(MulticancerSteinbock)
-export(available_10x_xen_zips)
+export(availableOSN)
 export(available_sdio)
-export(available_spd_zarr_zips)
-export(get_demo_SD)
-export(merfish_demo_path)
-export(path_to_10x_xen_demo)
-export(spd_demo_cached_path)
-export(spdzPath)
-export(unzip_merfish_demo)
-export(unzip_spd_demo)
+export(loadFromOSN)
 export(use_sdio)
 import(BiocFileCache)
-importFrom(basilisk,BasiliskEnvironment)
+importClassesFrom(SpatialData,SpatialData)
+importFrom(SpatialData,readSpatialData)
 importFrom(utils,unzip)

R/availableOSN.R

@@ -0,0 +1,23 @@
+#' use 'paws::s3' to interrogate an NSF Open Storage Network 
+#' bucket for zipped zarr archives for various platforms
+#' @examples
+#' if (requireNamespace("paws")) {
+#'   availableOSN()
+#' }
+#' @export
+availableOSN <- function() {
+    if (!requireNamespace("paws")) 
+        stop("install 'paws' to use this function; without it",
+            " we can't check existence of data in OSN bucket")
+#  x = curl::curl("https://mghp.osn.xsede.org/bir190004-bucket01")
+#  y = xml2::read_xml(x)
+#  z = xml2::as_list(y)
+    message("checking Bioconductor OSN bucket...")
+    s3 <- paws::s3(
+        credentials=list(anonymous=TRUE),
+        endpoint="https://mghp.osn.xsede.org")
+    zz <- s3$list_objects("bir190004-bucket01") 
+    allk <- lapply(zz$Contents, "[[", "Key")
+    basename(grep("BiocSpatialData\\/", allk, value=TRUE))
+}
+

R/get_demo_SD.R

@@ -11,66 +11,21 @@
 #' The individual functions in this package give similarly detailed references.
 "demo_spatialdata"
 
-#' @title retrieve scverse-curated `SpatialData` .zarr archive
-#' @aliases MouseIntestineVisHD
-#' 
-#' @description
-#' This function consolidates the retrieval and caching and transformation 
-#' of scverse-curated Zarr archives and 10x-curated Xenium archives.
-#' 
+#' all logic for finding, caching, loading an OSN-based dataset, hidden
+#' @importFrom SpatialData readSpatialData
+#' @importClassesFrom SpatialData SpatialData
 #' @param patt character(1) sufficient to identify an OSN resource
 #' @param cache like `BiocFileCache`
 #' @param target character(1), defaults to tempfile(); use a different 
 #'   value if you wish to retain the unzipped .zarr store persistently.
-#' 
-#' @details
-#' \describe{
-#' \item{
-#'   \code{MouseIntestineVisHD()}}{
-#'   Visium HD 3.0.0 (10x Genomics) dataset of mouse intestine; source:
-#' \emph{https://www.10xgenomics.com/datasets/visium-hd-cytassist-gene-expression-libraries-of-mouse-intestine}}
-#' \item{
-#'   \code{LungAdenocarcinomaMCMICRO()}}{
-#'   MCMICRO dataset of human small cell lung adenocarcinoma}
-#' \item{
-#'   \code{MouseBrainMERFISH()}}{
-#'   MERFISH dataset of mouse brain tissue}
-#' \item{
-#'   \code{MulticancerSteinbock()}}{
-#'   imaging mass cytometry dataset of four cancers; source:
-#'   \emph{https://www.nature.com/articles/s41596-023-00881-0}}
-#' \item{
-#'   \code{ColorectalCarcinomaMIBITOF()}}{
-#'   MIBI-TOF dataset of colorectal carcinoma}
-#' \item{
-#'   \code{JanesickBreastVisiumEnh()}}{
-#'   Visium (10x Genomics) dataset of breast cancer; source: 
-#'   \emph{https://www.nature.com/articles/s41467-023-43458-x}}
-#' \item{
-#'   \code{JanesickBreastXeniumRep1/2()}}{
-#'   two Xenium (10x Genomics) sections associated with
-#'   the above Visium section from Janesick \emph{et al.}}
-#' \item{
-#'   \code{Breast2fov_10x()}}{
-#'   Xenium (10x Genomics) data on breast cancer, trimmed to 2 FOVs; source: 
-#'   \emph{https://www.10xgenomics.com/support/software/xenium-onboard-analysis/latest/resources/xenium-example-data}}
-#' \item{
-#'   \code{Lung2fov_10x()}}{
-#'   Xenium (10x Genomics) data on lung cancer, trimmed to 2 FOVs; source: 
-#'   \emph{https://www.10xgenomics.com/support/software/xenium-onboard-analysis/latest/resources/xenium-example-data}}
-#' \item{
-#'   \code{HumanLungMulti_10x()}}{
-#'   Xenium (10x Genomics) data on lung cancer;
-#'   source: \emph{https://www.10xgenomics.com/datasets/preview-data-ffpe-human-lung-cancer-with-xenium-multimodal-cell-segmentation-1-standard}}
-#' }
-#' 
-#' @examples
-#' # the following are equivalent:
-#' get_demo_SD("merfish")
-#' MouseBrainMERFISH()
-#' 
-#' @export
-get_demo_SD <- function(patt, 
+#' @note This function checks for stale element in cache and uses bfcupdate to rectify
+#' before retrieving from cache.
+# @examples
+# # the following are equivalent:
+# get_demo_SD("merfish")
+# MouseBrainMERFISH()
+# 
+.get_demo_SD <- function(patt, 
     cache=BiocFileCache::BiocFileCache(),
     target=tempfile()) {
 
@@ -136,82 +91,139 @@ get_demo_SD <- function(patt,
     }
     # so a single pattern has hit, and it is in cache
     if (chkdf[ind,]$rname %in% xdzips) { # it is a Xenium 10x output resource
+        # check if update needed
+        stale = BiocFileCache::bfcneedsupdate(cache, chkdf[ind,]$rid)
+        if (stale) BiocFileCache::bfcupdate(cache, chkdf[ind,]$rid, fpath=chkdf[ind,]$fpath, rtype="web")
         preloc <- chkdf[ind,]$rpath
         td <- tempfile() # can't use target
         dir.create(td)
         unzip(preloc, exdir=td)  # manufacturer output
         use_sdio("xenium", srcdir=td, dest=target) # zarr in target
         return(SpatialData::readSpatialData(target))
     }
+# check again, this one is not in xdzips
+    stale = BiocFileCache::bfcneedsupdate(cache, chkdf[ind,]$rid)
+    if (stale) BiocFileCache::bfcupdate(cache, chkdf[ind,]$rid, fpath=chkdf[ind,]$fpath, rtype="web")
     loc <- chkdf[ind,]$rpath
     td <- target
     dir.create(td)
     unzip(loc, exdir=td)
     SpatialData::readSpatialData(dir(td, full.names=TRUE))
 }
 
-#' @rdname get_demo_SD
+#' @title retrieve scverse-curated `SpatialData` .zarr archive
+#' @aliases MouseIntestineVisHD
+#' 
+#' @description
+#' This function consolidates the retrieval and caching and transformation 
+#' of scverse-curated Zarr archives and 10x-curated Xenium archives.
+#' 
+#' @param target character(1), defaults to tempfile(); use a different 
+#'   value if you wish to retain the unzipped .zarr store persistently.
+#' 
+#' @details
+#' \describe{
+#' \item{
+#'   \code{MouseIntestineVisHD()}}{
+#'   Visium HD 3.0.0 (10x Genomics) dataset of mouse intestine; source:
+#' \emph{https://www.10xgenomics.com/datasets/visium-hd-cytassist-gene-expression-libraries-of-mouse-intestine}}
+#' \item{
+#'   \code{LungAdenocarcinomaMCMICRO()}}{
+#'   MCMICRO dataset of human small cell lung adenocarcinoma}
+#' \item{
+#'   \code{MouseBrainMERFISH()}}{
+#'   MERFISH dataset of mouse brain tissue}
+#' \item{
+#'   \code{MulticancerSteinbock()}}{
+#'   imaging mass cytometry dataset of four cancers; source:
+#'   \emph{https://www.nature.com/articles/s41596-023-00881-0}}
+#' \item{
+#'   \code{ColorectalCarcinomaMIBITOF()}}{
+#'   MIBI-TOF dataset of colorectal carcinoma}
+#' \item{
+#'   \code{JanesickBreastVisiumEnh()}}{
+#'   Visium (10x Genomics) dataset of breast cancer; source: 
+#'   \emph{https://www.nature.com/articles/s41467-023-43458-x}}
+#' \item{
+#'   \code{JanesickBreastXeniumRep1/2()}}{
+#'   two Xenium (10x Genomics) sections associated with
+#'   the above Visium section from Janesick \emph{et al.}}
+#' \item{
+#'   \code{Breast2fov_10x()}}{
+#'   Xenium (10x Genomics) data on breast cancer, trimmed to 2 FOVs; source: 
+#'   \emph{https://www.10xgenomics.com/support/software/xenium-onboard-analysis/latest/resources/xenium-example-data}}
+#' \item{
+#'   \code{Lung2fov_10x()}}{
+#'   Xenium (10x Genomics) data on lung cancer, trimmed to 2 FOVs; source: 
+#'   \emph{https://www.10xgenomics.com/support/software/xenium-onboard-analysis/latest/resources/xenium-example-data}}
+#' \item{
+#'   \code{HumanLungMulti_10x()}}{
+#'   Xenium (10x Genomics) data on lung cancer;
+#'   source: \emph{https://www.10xgenomics.com/datasets/preview-data-ffpe-human-lung-cancer-with-xenium-multimodal-cell-segmentation-1-standard}}
+#' }
+
+#' @rdname MouseIntestineVisHD
 #' @export
 MouseIntestineVisHD <- function(target=tempfile()) { 
-    get_demo_SD("visium_hd_3.0.0", target=target)
+    .get_demo_SD("visium_hd_3.0.0", target=target)
 }
 
-#' @rdname get_demo_SD
+#' @rdname MouseIntestineVisHD
 #' @export
 LungAdenocarcinomaMCMICRO <- function(target=tempfile()) {
-    get_demo_SD("mcmicro_io", target=target)
+    .get_demo_SD("mcmicro_io", target=target)
 }
 
-#' @rdname get_demo_SD
+#' @rdname MouseIntestineVisHD
 #' @export
 MouseBrainMERFISH = function(target=tempfile()) {
-    get_demo_SD("merfish", target=target)
+    .get_demo_SD("merfish", target=target)
 }
 
-#' @rdname get_demo_SD
+#' @rdname MouseIntestineVisHD
 #' @export
 MulticancerSteinbock <- function(target=tempfile()) {
-    get_demo_SD("steinbock_io", target=target)
+    .get_demo_SD("steinbock_io", target=target)
 }
 
-#' @rdname get_demo_SD
+#' @rdname MouseIntestineVisHD
 #' @export
 ColorectalCarcinomaMIBITOF <- function(target=tempfile()) {
-    get_demo_SD("mibitof", target=target)
+    .get_demo_SD("mibitof", target=target)
 }
 
-#' @rdname get_demo_SD
+#' @rdname MouseIntestineVisHD
 #' @export
 JanesickBreastVisiumEnh <- function(target=tempfile()) {
-    get_demo_SD("visium_associated_xenium_io", target=target)
+    .get_demo_SD("visium_associated_xenium_io", target=target)
 }
 
-#' @rdname get_demo_SD
+#' @rdname MouseIntestineVisHD
 #' @export
 JanesickBreastXeniumRep1 <- function(target=tempfile()) {
-    get_demo_SD("xenium_rep1_io", target=target)
+    .get_demo_SD("xenium_rep1_io", target=target)
 }
 
-#' @rdname get_demo_SD
+#' @rdname MouseIntestineVisHD
 #' @export
 JanesickBreastXeniumRep2 <- function(target=tempfile()) {
-    get_demo_SD("xenium_rep2_io", target=target)
+    .get_demo_SD("xenium_rep2_io", target=target)
 }
 
-#' @rdname get_demo_SD
+#' @rdname MouseIntestineVisHD
 #' @export
 Breast2fov_10x <- function(target=tempfile()) {
-    get_demo_SD("human_Breast_2fov", target=target)
+    .get_demo_SD("human_Breast_2fov", target=target)
 }
 
-#' @rdname get_demo_SD
+#' @rdname MouseIntestineVisHD
 #' @export
 Lung2fov_10x <- function(target=tempfile()) {
-    get_demo_SD("human_Lung_2fov", target=target)
+    .get_demo_SD("human_Lung_2fov", target=target)
 }
 
-#' @rdname get_demo_SD
+#' @rdname MouseIntestineVisHD
 #' @export
 HumanLungMulti_10x <- function(target=tempfile()) {
-    get_demo_SD("HuLungXenmulti", target=target)
-}
+    .get_demo_SD("HuLungXenmulti", target=target)
+}

R/loadFromOSN.R

@@ -0,0 +1,26 @@
+#' use a string to identify a resource in SpatialData format and
+#' ingest via SpatialData::readSpatialData
+#' @param stub character(1) a string that identifies a resource
+#' @return an instance of SpatialData, or NULL if the stub does not
+#' uniquely match (using grep()) the name of any resource
+#' @examples
+#' lu = loadFromOSN("Lung2")
+#' lu
+#' @export
+loadFromOSN = function(stub) { 
+  opts = ls("package:SpatialData.data")
+  hit = grep(stub, opts, value=TRUE)
+  if (!is.na(hit[1]) && length(hit)==1L) return(get(hit)())
+  else if (is.na(hit[1])) {
+    message("stub provided has no match in OSN resources")
+    message("returning NULL")
+    }
+  else {
+    message("stub does not uniquely match an OSN resource")
+    message("matched: ")
+    print(hit)
+    message("returning NULL")
+  }
+  NULL
+}
+