Run UMAP and save fgraph and embeddings in Seurat object

Usage

RunUMAPCustom(
  obj,
  reduction = "pca",
  dims = NULL,
  fgraph_only = FALSE,
  graph.name = NULL,
  reduction.name = "umap",
  assay = NULL,
  key = "UMAP_",
  n_neighbors = 30,
  n_components = 2,
  metric = "cosine",
  spread = 1,
  min_dist = 0.3,
  n_threads = NULL,
  fast_sgd = TRUE,
  verbose = TRUE,
  ...
)

Arguments

obj: A Seurat object.
reduction: Name of dimensional reduction in obj to use as input to UMAP.
dims: Dimensions of reduction to use as input to UMAP. If NULL, use all dimensions.
fgraph_only: If TRUE, only compute and store the fuzzy simplicial graph (fgraph) and skip UMAP embedding.
graph.name: Name of graph to store the fgraph in obj. Defaults to <assay>_fgraph.
reduction.name: Name of dimensional reduction to store the UMAP embeddings in obj. Defaults to "umap".
assay: Assay to set as default assay for the new dimensional reduction. If NULL, use the default assay of obj.
key: Key prefix to use for the new dimensional reduction. Defaults to "UMAP_".
n_neighbors: Number of nearest neighbors to use in UMAP. See ?uwot::umap for details.
n_components: Number of UMAP dimensions to compute. Ignored if fgraph_only is TRUE.
metric: Distance metric to use in UMAP. See ?uwot::umap for details.
spread: UMAP spread parameter. See ?uwot::umap for details.
min_dist: UMAP minimum distance parameter. See ?uwot::umap for details.
n_threads: Number of threads to use in UMAP. See ?uwot::umap for details.
fast_sgd: Whether to use the fast stochastic gradient descent optimization in UMAP. See ?uwot::umap for details.
verbose: Whether to print progress messages.
...: Additional parameters to pass to uwot::umap.

Value

The input Seurat object with the fgraph and (optionally) UMAP embeddings added.