Neighbour composition matrix

From a metadata file, given a list of cells with the corrisponding annotated cell type, build a neighbor matrix reporting the number of occurrences for each cell type

nnMat(
  data,
  label = NULL,
  method = c("Q", "C", "D", "K", "M"),
  layers = 1,
  snap = NULL,
  k = NULL,
  d.max = NULL,
  soi = F,
  ids_anno = NULL,
  return.seurat = T
)

Arguments

data: a Seurat object containing Kandinsky data
label: character string specifying the variable name to be used to define cell annotation groups
method: character string specifying the method to be used to identify neighbours. Must be one of the following: 'Q': queen contiguity method,check for contact (not overlap) between any edge or side od two polygons (refers to the queen movement rule in chess). Currently only applicable for Visium/Visium-HD data; 'C': centroid-based method, use maximum centroid distance threshold to identify spot/cell neighbours; 'D': Delaunay triangulation method; 'K': KNN method, define k closest neighbours to each spot/cell; 'M': membrane-based method, check for the occurrence of a physical contact/intersection within a distance threshold between cell boundaries. Not applicable in the case of Visium spots. When argument method is not specified or is set to NULL, the function will use nb slot already stored in Kandinsky object to create neighbour matrix
layers: numeric, number of concentric contiguous layers to include in spot neighbourhood. Only Applied when 'nb.method = Q'. Default is 1.
snap: numeric, extra distance accepted between polygon borders for contiguity relation. Applied when 'method' is set to Q. Can't work with point geometries like cell centroids.
k: numeric, number of nearest neighbours to be searched. Applied when method is set to K.
d.max: numeric, maximum centroid or membrane distance threshold. Applied when method is set to C or M.
soi: boolean, whether or not filter Delaunay network to keep sphere of influence (SOI) graph. Default is FALSE.
ids_anno: optional, character string specifying any extra variable to be added to the final distance matrix
return.seurat: boolean, whether returning neighbour matrix alone (FALSE) or the input Seurat object with a new nnMat slot as part of the Kandinsky data

Arguments

See also