Skip to contents

Read Visium HD data

Source: R/read.R
readVisiumHD.Rd

This function reads Visium HD Space Ranger output into R.

Usage

readVisiumHD(
  data_dir,
  bin_size = c(2L, 8L, 16L),
  sample_id = NULL,
  type = c("HDF5", "sparse"),
  data = c("filtered", "raw"),
  images = c("lowres", "hires"),
  unit = c("full_res_image_pixel", "micron"),
  style = "W",
  zero.policy = NULL,
  row.names = c("id", "symbol"),
  flip = c("geometry", "image"),
  add_graph = FALSE,
  rotate = FALSE
)

Arguments

data_dir

Directory

bin_size

One or more resolutions to load, must be 2, 8, or 16. Can be either integer or character.

sample_id

Which sample(s) in the SFE object to use for the graph. Can also be "all", which means this function will compute the graph for all samples independently.

type

Either "HDF5", and the matrix will be represented as TENxMatrix, or "sparse", and the matrix will be read as a dgCMatrix.

data

character string specifying whether to read in filtered (spots mapped to tissue) or raw data (all spots).

images

character vector specifying which images to include. Valid values are "lowres", "hires", "fullres", "detected", "aligned"

unit

Whether to use pixels in full resolution image or microns as the unit. If using microns, then spacing between spots in pixels will be used to convert the coordinates into microns, as the spacing is known to be 100 microns. This is used to plot scale bar.

style

style can take values “W”, “B”, “C”, “U”, “minmax” and “S”

zero.policy

default NULL, use global option value; if FALSE stop with error for any empty neighbour sets, if TRUE permit the weights list to be formed with zero-length weights vectors

row.names

String specifying whether to use Ensembl IDs ("ID") or gene symbols ("Symbol") as row names. If using symbols, the Ensembl ID will be appended to disambiguate in case the same symbol corresponds to multiple Ensembl IDs.

flip

Whether to flip the geometries or the images, because in sf and terra, the geometries use the Cartesian coordinates greater y coordinates going up, while in images, greater y values go down. Originally the Visium spots are in pixels in full res image. Either the image or the geometry needs to be flipped for them match in the Cartesian coordinate system.

add_graph

c(local), if to add spatial neighborhood graph for spots and only if c(data = "filtered"). Default is c(TRUE). This is optional because for larger datasets, the graph can take a while to compute.

rotate

Logical, whether to rotate the geometry, because usually the grid of spots is slightly, but just very slightly, rotated from the perfect horizontal line. The spots can be rotated so the square polygons are more accurate, because for computational efficiency, st_buffer is used to create the polygons which are not rotated. This is optional because the rotation is very slight.

Value

An SFE object if `length(bin_size) == 1L`, otherwise a list of SFE objects each element of which is for one bin size. They're not concatenated since it might not make sense to perform joint analyses on the different resolutions that benefit from having them in the same SFE object, unlike different biological replica. Here unlike in read10xVisiumSFE, the centroids geometry is also added because it will greatly facilitate plotting when there are many spots when not zooming in. See the scattermore argument in plotSpatialFeature.

Examples

#