Calculates PCA on Pearson residuals for single cell — rs_sc_pca

Runs the PCA on the residuals a fitted model implies, rather than on the stored normalised layer. The residual columns are dense by construction, since a zero count still has a residual, so there is no sparse or streaming variant of this path.

Two settings are refused rather than ignored: the PFlogPF transform, which belongs to the normalised layer, and variance normalisation, which would flatten the very ranking the residuals produce.

Usage

rs_sc_pca_residuals(
  f_path_gene,
  residual_fit,
  no_pcs,
  pca_params,
  cell_indices,
  gene_indices,
  seed,
  return_scaled,
  verbose
)

Arguments

f_path_gene: String. Path to the counts_genes.bin file.
residual_fit: List. A fit from rs_sc_fit_residuals().
no_pcs: Integer. Number of PCs to calculate.
pca_params: Named list. Contains the parameters to use for this PCA run. clr and normalise_variance must both be FALSE.
cell_indices: Integer vector. The cell indices to use. (0-indexed!) Must be the selection the fit was fitted on.
gene_indices: Integer vector. The gene indices to use. (0-indexed!) Every one must be covered by the fit.
seed: Integer. Random seed for the randomised SVD.
return_scaled: Boolean. Shall the scaled data be returned.
verbose: Integer. 0L - quiet; 1L - normal verbosity; 2L - detailed verbosity.

Value

A list with the following items

scores - The samples projected on the PCA space.
loadings - The loadings of the features for the PCA.
singular_values - The singular values for the PCA.
scaled - The scaled matrix if return_scaled = TRUE, otherwise NULL.