scDHA: scDHA

In scDHA: Single-Cell Decomposition using Hierarchical Autoencoder

scDHA

Description

The main function to perform dimension deduction and clustering.

Usage

scDHA(
  data = data,
  k = NULL,
  method = "scDHA",
  sparse = FALSE,
  n = 5000,
  ncores = 10L,
  gen_fil = TRUE,
  do.clus = TRUE,
  sample.prob = NULL,
  seed = NULL
)

Arguments

data	Gene expression matrix, with rows represent samples and columns represent genes.
k	Number of clusters, leave as default for auto detection. Has no effect when do.clus = False.
method	Method used for clustering. It can be "scDHA" or "louvain". The default setting is "scDHA".
sparse	Boolen variable indicating whether data is a sparse matrix. The input must be a non negative sparse matrix.
n	Number of genes to keep after feature selection step.
ncores	Number of processor cores to use.
gen_fil	Boolean variable indicating whether to perform scDHA gene filtering before performing dimension deduction and clustering.
do.clus	Boolean variable indicating whether to perform scDHA clustering. If do.clus = False, only dimension deduction is performed.
sample.prob	Probability used for classification application only. Leave this parameter as default, no user input is required.
seed	Seed for reproducibility.

Value

List with the following keys:

• cluster - A numeric vector containing cluster assignment for each sample. If do.clus = False, this values is always NULL.

• latent - A matrix representing compressed data from the input data, with rows represent samples and columns represent latent variables.

Examples

library(scDHA)
#Load example data (Goolam dataset)
data('Goolam'); data <- t(Goolam$data); label <- as.character(Goolam$label)
#Log transform the data 
data <- log2(data + 1)
if(torch::torch_is_installed()) #scDHA need libtorch installed
{
  #Generate clustering result, the input matrix has rows as samples and columns as genes
  result <- scDHA(data, ncores = 2, seed = 1)
  #The clustering result can be found here 
  cluster <- result$cluster
}

scDHA documentation built on May 29, 2024, 4:51 a.m.