Performs dimension reduction and adds results to the given input.

See details for available dimension reduction techniques that can be applied to the object.

reduce_dimension(object, flavor = c("umap", "som", "asvd", "isomds",
  "rsvd"), slot = flavor, ...)

Arguments

object	A SingleCellExperiment object.
flavor	Determines which dimension reduction technique to apply.
slot	Determines which entry of the reducedDims slot to use for reduced embedding.
...	Additional parameters passed to functions.

Value

A SingleCellExperiment object with modified reducedDims slot.

Details

The following dimension reduction techniques are available:

• UMAP - Uniform Manifold Approximation and Projection, see umap.

• approximate SVD - see approximate_svd

• randomized SVD - see randomized_svd

• Multidimensional scaling - see isomds

• Self-organizing maps - see scSOM

Examples

# NOT RUN {
# Apply UMAP on normalized expression values, but store results in a slot called dimred_umap.
obj <- reduce_dimension(obj, flavor = "umap", slot = "dimred_umap", exprs_values = "norm_exprs")

# Apply randomized SVD on normalized expression values, keeping 50 dimensions, followed by umap.
# Using magrittr pipes, the result of randomized SVD is stored in the rsvd slot and
# picked up by umap.
obj %<>%
  reduce_dimension(flavor = "rsvd", exprs_values = "norm_exprs", n_dims = 50) %>%
  reduce_dimension(flavor = "umap", exprs_values = NULL, use_dimred = "rsvd")
# }