Sorts cell metadata variable by similarity using hierarchical clustering

Compute distance matrix from a feature/variable matrix and perform hierarchical clustering to order variables (for example, cell types) according to their similarity.

SortIdents(
  object,
  layer = "data",
  assay = NULL,
  label = NULL,
  dendrogram = FALSE,
  method = "euclidean",
  verbose = TRUE
)

Arguments

object

A Seurat object containing single-cell data.

layer

The layer of the data to use (default is "data").

assay

Name of assay to use. If NULL, use the default assay

label

Metadata attribute to sort. If NULL, uses the active identities.

dendrogram

Logical, whether to plot the dendrogram (default is FALSE).

method

The distance method to use for hierarchical clustering (default is 'euclidean', other options from dist are 'maximum', 'manhattan', 'canberra', 'binary' and 'minkowski').

verbose

Display messages

Value

The Seurat object with metadata variable reordered by similarity. If the metadata variable was a character vector, it will be converted to a factor and the factor levels set according to the similarity ordering. If active identities were used (label=NULL), the levels will be updated according to similarity ordering.

Examples

atac_small$test <- sample(1:10, ncol(atac_small), replace = TRUE)
atac_small <- SortIdents(object = atac_small, label = 'test')
#> Creating pseudobulk profiles for 10 variables across 323 features
#> Computing euclidean distance between pseudobulk profiles
#> Clustering distance matrix
print(levels(atac_small$test))
#>  [1] "8"  "2"  "9"  "3"  "4"  "10" "7"  "1"  "5"  "6"