visiumSpotBins: Build spot-level hexagonal bins from a Visium...

In blisa: Infer Cell-Cell Communication from Spatial Transcriptomics

Build spot-level hexagonal bins from a Visium SpatialExperiment

Description

Turns a 10x Visium SpatialExperiment (as produced by scider::readVisium()) into the pre-binned inputs expected by blisa.default, treating each Visium spot as one bin. The spot coordinates are rebuilt from the integer array_col/array_row indices onto an exact, de-tilted hexagonal lattice with 100 um (spot_pitch) spacing – the standalone equivalent of scider::realignVisium().

Usage

visiumSpotBins(
  spe,
  spot_pitch = NULL,
  in_tissue_only = TRUE,
  hexagons = TRUE,
  min_total_counts = 10,
  verbose = FALSE
)

Arguments

spe	A Visium SpatialExperiment with a "counts" assay whose row names are gene symbols. array_col/array_row in colData are used to rebuild the exact lattice when present; otherwise spatialCoords(spe) is used as-is.
spot_pitch	Numeric or NULL. Spot center-to-center spacing used as bin_size. NULL (default) means: use 100 um for the array-index lattice, or measure it from the coordinates in the fallback. Supply a number to force it (interpreted in coordinate units).
in_tissue_only	Logical. Keep only spots with in_tissue == 1. Default TRUE.
hexagons	Logical. If TRUE (default), build hexagonal bin polygons; if FALSE, use point geometry (faster, but plotHotspots then draws points).
min_total_counts	Numeric. Drop spots whose total counts fall below this threshold. Default 10. Set to 0 to keep all spots.
verbose	Logical. Print progress messages. Default FALSE.

Details

On Visium, each spot is already a spatial bin, so no cell-to-bin aggregation is needed (unlike hexBinCells). Because array_col increments by 2 within a row and alternate rows shift array_col by 1, the half-row stagger is already encoded in the indices; the lattice therefore needs no odd/even phase correction. Only the bin centroids are used by computeSpatialWeights; the hexagon polygons are for plotting (they do tessellate the lattice exactly).

When array_col/array_row are absent (e.g. a SpatialExperiment not built by scider::readVisium()), the function falls back to using spatialCoords(spe) directly: spots are still one bin each, but the de-tilt/exact-lattice step is skipped. In that case, if spot_pitch is NULL, the pitch is measured from the coordinates as the median nearest-neighbour distance. Since the physical Visium pitch is 100 um, 100 / pitch is microns-per-unit; the coordinates are rescaled to microns by that factor (near-identity when they are already microns), so the returned pitch is 100 and downstream bin_size/dmax are in microns. If you instead supply spot_pitch, no rescaling is done and everything stays in the coordinates' native units.

Value

A list with:

counts_matrix

Gene-by-spot count matrix (columns match bins).

bins

An sf object of per-spot bins with bin_id, n_cells (always 1), total_counts, and img_x/ img_y (the raw image-registered spot coordinates, for overlaying plots on the H&E image) columns. Row order matches the columns of counts_matrix.

pitch

The spot spacing used (= the recommended bin_size).

coord_unit

"micron", "pixel/other", or NA (when spot_pitch was supplied) – the inferred coordinate unit.

See Also

blisa, hexBinCells

blisa documentation built on Oct. 30, 2026, 9:10 a.m.