SilhouetteBoot: Bootstrapped silhouette score

In RuiyuRayWang/scWidgets: A collection of scRNAseq analysis functions

Description

Unsupervised way to choose the optimal clustering resolutions or number of clusters. Perform bootstrapping (repeated random subsampling with replacement) to achieve better estimation of Silhouette score.

Usage

SilhouetteBoot(
  object,
  cluster_ident,
  data_from = "pca",
  balanced = TRUE,
  iteration = 100,
  subsample.size = NULL,
  replace = TRUE,
  verbose = TRUE
)

Arguments

object	Seurat object.
cluster_ident	Identity class of clustering.
data_from	By default, data_from = "pca", extract pca embedding as input data matrix. In case of data_from = "expression", extract data from original expression matrix.
balanced	Whether or not to perform balanced subsampling, which selects proportional size of cells from each cluster. Default is TRUE.
iteration	Number of iterations for bootstrapping. Default = 100.
subsample.size	Sample size for bootstrapping. If unspecified, subsample one fifth (1/5) of the total population.
replace	Sampling with or without replacement. Default is TRUE.
verbose	Display progress information.

RuiyuRayWang/scWidgets documentation built on Dec. 18, 2021, 11:54 a.m.