Cluster neighborhoods — neighborhoodClusters • SpatialMap

Assuming that cellNeighborhoods has been run, clusters the neighborhoods

Usage

neighborhoodClusters(
  object,
  nn,
  feature,
  cluster = "km",
  k = NULL,
  col.name = NULL,
  quietly = F
)

Arguments

object: A SpatialMap object
nn: A NN graph name
feature: The name of the cellMetadata feature (e.g. cell type) used for defining neighborhoods.
cluster: How to cluster the cells' local neighbor composition vectors to create neighborhoods. Currently only k-means is supported ("km").
k: How many clusters to produce using k-means on the neighborhood counts matrix. If NULL (default), then the function will optimize k using the silhouette score. Will also plot the silhouette scores for each tested value of k as a side effect.
col.name: The desired name of the metadata column for neighborhood clusters. Defaults to something descriptive
quietly: If true, will not print output messages

Value

A SpatialMap object with new cell neighborhood memberships added to the cellMetadata slot as a factor. The new column in cellMetadata will be named according to the value passed to col.name, or will be named by concatenating "nh_{nn}_{feature}".

Details

See vignette("AnalysisGuide6_Neighborhoods") for a representative workflow that uses this function.