Skip to contents

Run t-SNE on a SingleCells object

tsne_sc.Rd

Wrapper around manifoldsR::tsne() for the SingleCells class.

Usage

tsne_sc(
  object,
  use_knn = TRUE,
  embd_to_use = "pca",
  no_embd_to_use = NULL,
  n_dim = 2L,
  perplexity = 30,
  approx_type = c("bh", "fft"),
  knn_method = c("hnsw", "balltree", "annoy", "nndescent", "exhaustive"),
  nn_params = manifoldsR::params_nn(),
  tsne_params = manifoldsR::params_tsne(),
  seed = 42L,
  .verbose = TRUE
)

Arguments

object

SingleCells class.

use_knn

Boolean. Use the kNN graph found in the object. Defaults to TRUE. If not available, will default to the embedding.

embd_to_use

String. The embedding to use for t-SNE. Must be available in the object.

no_embd_to_use

Optional integer. Number of embedding dimensions to use. If NULL all will be used.

n_dim

Integer. Number of t-SNE dimensions. Currently only 2L is supported. Defaults to 2L.

perplexity

Numeric. Perplexity parameter. Typical values between 5 and 50. Defaults to 30.0.

approx_type

String. Approximation method. One of "bh" (Barnes-Hut) or "fft". Defaults to "bh".

knn_method

String. Approximate nearest neighbour algorithm. One of "hnsw", "balltree", "annoy", "nndescent", or "exhaustive".

nn_params

Named list. See manifoldsR::params_nn().

tsne_params

Named list. See manifoldsR::params_tsne().

seed

Integer. For reproducibility.

.verbose

Boolean. Controls verbosity.

Value

The object with a "tsne" embedding added.