Skip to contents

Run PCA for single cell

calculate_pca_sc.Rd

This function will run PCA on the detected highly variable genes. You can use randomised SVD for speed and there is an option for sparse SVD for very large data sets to avoid memory pressure.

Usage

calculate_pca_sc(
  object,
  no_pcs,
  pca_params = params_sc_pca(),
  sparse_svd = FALSE,
  hvg = NULL,
  seed = 42L,
  residuals = FALSE,
  .verbose = TRUE
)

Arguments

object

SingleCells, MetaCells (or potentially other) class.

no_pcs

Integer. Number of PCs to calculate.

pca_params

Named list. Controls the parameters to be used for the PCA calculation which is single cell-specific, see params_sc_pca()

sparse_svd

Boolean. Shall sparse solvers be used that do not do scaling. If set to yes, in the case of random_svd = FALSE, Lanczos iterations are used to solve the sparse SVD. With random_svd = TRUE, the sparse initial matrix is multiplied with the random matrix, yielding a much smaller dense matrix that does not increase the memory pressure massively. Not used for MetaCells.

hvg

Optional integer. If you want to provide your own HVG genes. Otherwise, the function will default to what is found in get_hvg(). Please provide 1-indexed genes here! If you provide these, the internal HVG will be overwritten.

seed

Integer. Controls reproducibility. Only relevant if randomised_svd = TRUE.

residuals

Boolean. Run the PCA on the Pearson residuals of a model fitted with fit_residuals_sc() instead of the stored normalised layer. Needs params_sc_pca(normalise_variance = FALSE, clr = FALSE), since the residuals already carry the signal as variance, and does not support sparse_svd: a residual column is dense even where the counts are not. Not supported for MetaCells.

.verbose

Boolean or integer. Controls verbosity and returns run times. FALSE -> quiet, TRUE or 1L -> normal verbosity, 2L -> detailed verbosity.

Value

The function will add the PCA factors, loadings and singular values to the object cache in memory.

Examples

# PCA on the highly variable genes
sc <- demo_single_cells(prepped = FALSE)
sc <- find_hvg_sc(sc, hvg_no = 30L, .verbose = FALSE)
sc <- calculate_pca_sc(sc, no_pcs = 10L, .verbose = FALSE)
dim(get_pca_factors(sc))
#> [1] 500  10

unlink(sc@dir_data, recursive = TRUE, force = TRUE)