Build a sparse neighbor-weight matrix via k-NN, optionally filtered by radius
Usage
buildKNNWeights(
ref_coords,
query_coords = ref_coords,
k = 10,
radius = Inf,
include.self = TRUE
)Arguments
- ref_coords
data.frame/matrix of coordinates for the *source* units (rows = e.g. cells or spots).
- query_coords
data.frame/matrix of coordinates for the *output* units (rows = e.g. spatial neighborhoods or spots), columns X, Y. Defaults to `ref_coords`.
- k
number of nearest neighbors to search per query point.
- radius
numeric distance cutoff; neighbors farther than this are dropped. Use `Inf` (default) to keep all `k` neighbors regardless of distance.
- include.self
a boolean specifying whether to include self from the knn.