Add support for seurat hdf5 format.

R/H5SparseMatrixSeed-class.R

@@ -136,8 +136,12 @@ read_h5sparse_component <- function(filepath, group, name,
         ## 10x format
         return(read_h5sparse_component(filepath, group, "shape"))
     }
-    ## h5ad format
     h5attrs <- h5readAttributes(filepath, group)
+    ## h5seurat format
+    shape <- h5attrs$dims
+    if(!is.null(shape))
+        return(as.vector(shape))
+    ## h5ad format
     shape <- h5attrs$shape
     if (is.null(shape))
         shape <- h5attrs$h5sparse_shape
@@ -157,8 +161,12 @@ read_h5sparse_component <- function(filepath, group, name,
         ## 10x format
         return("csr")
     }
-    ## h5ad format
     h5attrs <- h5readAttributes(filepath, group)
+    ## h5seurat format
+    shape <- h5attrs$dims
+    if(!is.null(shape))
+        return("csr")
+    ## h5ad format
     h5sparse_layout <- h5attrs[["encoding-type"]]
     if (is.null(h5sparse_layout))
         h5sparse_layout <- h5attrs[["h5sparse_format"]]

R/TENxMatrixSeed-class.R

@@ -12,19 +12,31 @@ setClass("TENxMatrixSeed", contains="CSC_H5SparseMatrixSeed")
 
 .find_rownames_dataset <- function(filepath, group)
 {
-    name <- "genes"
-    if (h5exists(filepath, paste(group, name, sep="/")))
-        return(name)
-    name <- "features/id"
-    if (h5exists(filepath, paste(group, name, sep="/")))
-        return(name)
+    # features is for h5seurat.
+    candidate <- c("genes", "features", "features/id")
+    for(i in candidate)
+    {
+        if (h5exists(filepath, file.path(group, i)))
+            return(i)
+    }
+    NULL
+}
+
+.read_h5seurat_component <- function(filepath, name){
+    if (h5exists(filepath, name))
+        return(as.vector(h5mread(filepath, name)))
     NULL
 }
 
 ### Return the rownames of the matrix.
 .load_tenx_rownames <- function(filepath, group)
 {
     name <- .find_rownames_dataset(filepath, group)
+    # h5seurat data location: /path/to/group/../features
+    # Find dataset in altered group
+    if (is.null(name)){
+        group <- gsub('/[^/]+$','',group)
+        name <- .find_rownames_dataset(filepath, group)}
     if (is.null(name))
         return(NULL)
     read_h5sparse_component(filepath, group, name)
@@ -33,9 +45,12 @@ setClass("TENxMatrixSeed", contains="CSC_H5SparseMatrixSeed")
 ### Return the colnames of the matrix.
 .load_tenx_barcodes <- function(filepath, group)
 {
-    if (!h5exists(filepath, paste0(group, "/barcodes")))
-        return(NULL)
-    read_h5sparse_component(filepath, group, "barcodes")
+    if (h5exists(filepath, paste0(group, "/barcodes")))
+        return(read_h5sparse_component(filepath, group, "barcodes"))
+    # h5seurat
+    if (h5exists(filepath, "cell.names"))
+        return(read_h5sparse_component(filepath, '', "cell.names"))
+    NULL
 }
 
 

man/TENxMatrix-class.Rd

@@ -231,6 +231,20 @@ n_exprs3 <- lengths(nonzeros)
 ## Sanity checks:
 stopifnot(all.equal(lib_sizes, lib_sizes3))
 stopifnot(all.equal(n_exprs, n_exprs3))
+
+## ---------------------------------------------------------------------
+## SOME EXAMPLES OF H5SEURAT FORMAT
+## ---------------------------------------------------------------------
+fpath <- system.file('extdata','toy.h5seurat',package="HDF5Array")
+# Read count data of h5seurat dataset.
+group <- '/assays/RNA/counts'
+df.count <- HDF5Array::TENxMatrix(fpath, group)
+df.count[1:10,2:9]
+
+# Read normalized data of h5seurat dataset.
+group <- '/assays/RNA/data'
+df.data <- HDF5Array::TENxMatrix(fpath, group)
+df.data[1:10,2:9]
 }
 \keyword{classes}
 \keyword{methods}