Skip to contents

Run Seurat rPCA integration

seurat_rpca_sc.Rd

This function implements the reciprocal PCA (rPCA) anchor integration from Stuart, et al. It runs the same anchor pipeline as seurat_cca_sc() but builds a cheaper per-pair anchor space: each batch keeps its own PCA basis and the other batch's HVG expression is projected into it. Cross-batch mutual nearest neighbours are then found in these projected bases.

rPCA is faster than CCA and corrects less aggressively, which makes it the safer choice when batches share most of their cell types. As in Seurat, no gene-space anchor filter is applied, that step is CCA-only.

Usage

seurat_rpca_sc(
  object,
  batch_column,
  batch_hvg_genes,
  rpca_params = params_sc_seurat_rpca(),
  use_precomputed_pca = FALSE,
  seed = 42L,
  .verbose = TRUE
)

Arguments

object

SingleCells class.

batch_column

String. The column with the batch information in the obs data of the class.

batch_hvg_genes

Integer vector. These are the highly variable genes, identified by a batch-aware method. Please refer to find_hvg_batch_aware_sc() for more details. These genes have to be 0-indexed!

rpca_params

A list, please see params_sc_seurat_rpca(). The list has the following parameters:

  • dims - Integer. Dimensions used for the per-batch projections, the anchor kNN queries and the size of the returned embedding.

  • k_anchor - Integer. Neighbourhood size for the anchor search.

  • k_score - Integer. Neighbourhood size for the anchor scoring.

  • k_weight - Integer. Neighbourhood size for the kernel weights.

  • l2_norm - Logical. L2-normalise the projected embeddings per cell.

  • sd - Numeric. Bandwidth divisor of the Gaussian kernel.

  • knn - List of kNN parameters. See params_knn_defaults() for available parameters and their defaults.

  • pca - List of PCA parameters, see params_sc_pca() for available parameters and their defaults.

use_precomputed_pca

Boolean. Should the PCA in the object be used if found. This only applies to the union PCA that gets corrected, the per-batch PCAs are always recomputed because rPCA needs them.

seed

Integer. Random seed.

.verbose

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

Value

The object with the added "rpca" embedding.

References

Stuart, et al., Cell, 2019

Examples

# reciprocal PCA anchor integration, the cheaper sibling of CCA
sc <- demo_single_cells(
  syn_data_params = params_sc_synthetic_data(
    n_cells = 600L, n_genes = 50L, n_batches = 3L
  )
)
hvg <- find_hvg_batch_aware_sc(
  sc, hvg_no = 20L, batch_column = "batch_index", .verbose = FALSE
)
sc <- seurat_rpca_sc(
  sc,
  batch_column = "batch_index",
  batch_hvg_genes = hvg$hvg_gene_idx,
  rpca_params = params_sc_seurat_rpca(dims = 10L),
  .verbose = FALSE
)
dim(get_embedding(sc, "rpca"))
#> [1] 600  10

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