Skip to contents

Wrapper function for the Seurat rPCA parameters

params_sc_seurat_rpca.Rd

Wrapper function for the Seurat rPCA parameters

Usage

params_sc_seurat_rpca(
  dims = 30L,
  k_anchor = 5L,
  k_score = 30L,
  k_weight = 100L,
  l2_norm = TRUE,
  sd = 1,
  knn = list(),
  pca = params_sc_pca()
)

Arguments

dims

Integer. Number of dimensions used for the per-batch PCA projections, the anchor kNN queries and the size of the returned embedding. Defaults to 30L.

k_anchor

Integer. Neighbourhood size for the mutual nearest neighbour anchor search. Defaults to 5L.

k_score

Integer. Neighbourhood size for the shared-neighbour anchor scoring. Defaults to 30L.

k_weight

Integer. Neighbourhood size for the kernel weights applied during the correction. Defaults to 100L.

l2_norm

Boolean. Shall the projected embeddings be L2-normalised per cell. Defaults to TRUE.

sd

Numeric. Bandwidth divisor of the Gaussian kernel used for the anchor weights. Defaults to 1.0.

knn

List. Optional overrides for kNN parameters. See params_knn_defaults() for available parameters: k, knn_method, ann_dist, search_budget, n_trees, delta, diversify_prob, ef_budget, extract_knn, m, ef_construction, ef_search, n_list and n_probe. Note that k is unused here, the neighbourhood sizes come from k_anchor, k_score and k_weight. See params_knn_defaults() for the available elements. Defaults to list().

pca

List. Parameters to feed through to the optional recalculation of the PCA, see params_sc_pca(). See params_sc_pca() for the available elements. Defaults to params_sc_pca().

Value

A named list with the following elements:

  • dims - Integer. Number of dimensions used for the per-batch PCA projections, the anchor kNN queries and the size of the returned embedding. Defaults to 30L.

  • k_anchor - Integer. Neighbourhood size for the mutual nearest neighbour anchor search. Defaults to 5L.

  • k_score - Integer. Neighbourhood size for the shared-neighbour anchor scoring. Defaults to 30L.

  • k_weight - Integer. Neighbourhood size for the kernel weights applied during the correction. Defaults to 100L.

  • l2_norm - Boolean. Shall the projected embeddings be L2-normalised per cell. Defaults to TRUE.

  • sd - Numeric. Bandwidth divisor of the Gaussian kernel used for the anchor weights. Defaults to 1.0.

  • The elements of params_knn_defaults(), overridden by knn, spliced in at this position.

  • The elements of params_sc_pca(), overridden by pca, spliced in at this position.

References

Stuart, et al., Cell, 2019