Generate 2D cell embeddings using UMAP

calc_umap(fsce, expt = "rnaseq", method = "PCA", n_dims = NULL,
  n_neighbors = 30, min_dist = 0.3, metric = "euclidean",
  seed = NA, ...)

Arguments

fsce

FunctionalSingleCellExperiment object

expt

Data to use for calculating variable features (default is rnaseq). Must be present in names(fsce).

method

dimenality reduction method to use for UMAP (default is "PCA")

n_dims

number of dimensions to pass to UMAP, defaults to all present in dr matrix

n_neighbors

number of nearest neighbors to use for learning the manifold. Low values will preserve local structure, at the expense missing higher order organization. Higher values will capture more global structure but miss fine grained detail. Defaults to 30.

min_dist

Numeric between 0 and 0.99. min_dist controls how tightly points can be packed together in 2D space. Lower values will generate more clumpy projections, but more accurately preserve local structure.

metric

distance metric for UMAP, defaults to pearson.

seed

seed to generate reproducible UMAP projection. Defaults to no seed.

...

additional arguments for umap::umap()

Value

fsce with UMAP values added to reducedDims

See also

See https://umap-learn.readthedocs.io for a detailed description of parameters.