cluster.divisive: Fast divisive clustering of columns in a large sparse matrix

In zdebruine/scNMF: Fast Non-negative matrix factorization toolkit for single cell data

Description

Fast divisive clustering of columns in a large sparse matrix

Usage

cluster.divisive(
  input,
  min.samples = 10,
  subsample = 10000,
  seed = NULL,
  verbose = 2,
  min.dist = NULL,
  idents.prefix = "",
  average.expression = TRUE,
  details = FALSE,
  reduction.name = "dclus"
)

Arguments

input	A dgCMatrix matrix or a Seurat object with counts for clustering in the default assay counts slot
min.samples	The minimum number of samples permitted in a cluster. Real valued between 2 and ncol(input), defaults to 10
subsample	Number of samples to subsample for SVD and delta distance calculations, usually only useful for large matrices (> 10000 samples)
seed	Random seed for subsampling (NULL by default)
verbose	0 = no output, 1 = output for each generation, 2 = progress bar for each generation, 3 = details for each division
min.dist	Minimum delta distance, which is a proportion of how much more similar samples are to their assigned cluster than their opponent cluster. NULL by default. Values which approximate Newman-Girvan modularity of 0 are usually between 0.01 and 0.05.
idents.prefix	Prefix for ident names, useful to prevent later runs of cluster.divisive on the same object from overwriting previous idents
average.expression	Average expression of cluster centers and stash results in a dimensional reduction slot
details	Boolean, calculate additional statistics about each cluster (default is FALSE)
reduction.name	Name of \@reduction object if average.expression = TRUE

Value

If a Seurat object was input, a Seurat object with idents set to cluster IDs. If a sparse matrix was input, a list of cluster assignments for each sample.

zdebruine/scNMF documentation built on Jan. 1, 2021, 1:50 p.m.