Perform Clustering on a SingleCellExperiment object

Usage

perform_clustering(
  sce,
  ret_obj = FALSE,
  normalized_method = "LogNormalize",
  use.kmeans = FALSE,
  kmeans.nstart = 1,
  resolution = 0.5,
  step = 0.05,
  maxiter = 40,
  louvain.nstart = 1,
  louvain.alg = 1,
  seed.cluster = 0,
  kmeans.whiten = FALSE,
  pca.whiten = FALSE,
  npcs = 10,
  k.param = 20,
  test.use = "t",
  test.from.raw = FALSE,
  signal_measurement = "tstat",
  verbose = FALSE
)

Arguments

sce: A SingleCellExperiment object
ret_obj: (Default: FALSE) whether to return the sce after operation instead of returning the signal measurement
normalized_method: Normalization method. Possible choices: LogNormalize, sct, or none
use.kmeans: if TRUE, then clustering using kmeans with two clusters, otherwise, use Seurat::FindClusters and need to find resolution to achieve two clusters
kmeans.nstart: the nstart parameter for kmeans
resolution: the resolution parameter for Seurat::FindClusters
step: To find a resolution with desired number of clusters, the step size for changing the resolution parameter
maxiter: maximum iteration for searching resolution to have two clusters
louvain.nstart: the n.start parameter for Seurat::FindClusters
louvain.alg: the algorithm parameter for Seurat::FindClusters
seed.cluster: the random.seed parameter for Seurat::FindClusters
kmeans.whiten: whether to whitening for the kmeans clustering
pca.whiten: whether to whitening for PCA if using Louvain algorithms
npcs: number of PCs, the parameter npcs used in Seurat::RunPCA
k.param: the parameter k.param used in Seurat::FindNeighbors
test.use: The hypothesis testing for DE test. Possible choices: t, wilcox, poisson, and negbinom
test.from.raw: whether to perform the testing on the original data without normalization or on the normalized data
signal_measurement: the signal measurement. Possible choices: the test statistic tstat, or the signed p-value pval
verbose: whether to print internal log messages

Value

a vector of signal measurements for each feature