Scale each feature before embedding or clustering

One implementation for run_umap() and run_unsupervised_clusters(), so clusters and coordinates are computed in the same space.

Usage

scale_features(mat, method = "range")

Arguments

mat

numeric matrix, cells x features.

method

"none", "range" (0.5th-99.5th percentile range), "robust" (median/MAD) or "zscore".

Value

list(mat, params); params carries center and scale so a saved model can be applied to later data in the same space, or NULL for "none".