hippo: HIPPO's hierarchical clustering

In tk382/HIPPO: Heterogeneity-Induced Pre-Processing tOol

Description

HIPPO's hierarchical clustering

Usage

hippo(
  sce,
  K = 20,
  feature_method = c("zero_inflation", "deviance"),
  clustering_method = c("kmeans", "Seurat", "SC3"),
  z_threshold = 1.5,
  deviance_threshold = 150,
  outlier_proportion = 0.001,
  km_num_embeds = 10,
  km_nstart = 50,
  km_iter.max = 50,
  sc3_n_cores = NA,
  verbose = TRUE
)

Arguments

sce	SingleCellExperiment object
K	maximum number of clusters
feature_method	string, either "zero-inflation" or "deviance"
clustering_method	string, one of "kmeans", "Seurat", and "SC3"
z_threshold	numeric > 0 as a z-value threshold for selecting the features
deviance_threshold	numeric > 0 as a deviance threshold for selecting the features when method is "deviance
outlier_proportion	numeric between 0 and 1, a cut-off so that when the proportion of important features reach this number, the clustering terminates
km_num_embeds	number of cell embeddings to use in dimension reduction
km_nstart	number of tries for k-means for reliability
km_iter.max	number of maximum iterations for kmeans
sc3_n_cores	number of cores to use if your method is "SC3"
verbose	if set to TRUE, shows progress of the algorithm

Value

a list of clustering result for each level of k=1, 2, ... K.

Examples

data(toydata)
toydata = hippo(toydata,
          feature_method = "zero_inflation",
          clustering_method = "kmeans",
          K = 4,
          outlier_proportion = 0.00001)

tk382/HIPPO documentation built on Aug. 17, 2021, 2:29 p.m.