Skip to contents

project_pca

Source: R/data_preparation.R
project_PCA.Rd

Project Training and Test Seurat Objects onto Principal Components

This function projects both training and test Seurat objects onto a set of principal components derived from the training data.

Usage

project_pca(
  train_seurat,
  test_seurat,
  train_with_pcs,
  clust_pcs,
  dtype,
  verbose = 0,
  compute_train_eval = FALSE
)

Arguments

train_seurat

A Seurat object representing the training data set.

test_seurat

A Seurat object representing the test data set.

train_with_pcs

Which reduction should be used for training "odd_pca" or "even_pca"

clust_pcs

Which reduction was used for clustering

dtype

Type of data in the Seurat object "scRNA" or "CyTOF"

verbose

Integer verbosity level (0 = silent, 1 = milestones, 2 = detailed, 3 = includes Seurat output)

compute_train_eval

Logical; if TRUE, also compute the training data projected onto clustering PCs (train_proj_clust_pcs). Default is FALSE because this projection is not used in the standard clust_opt pipeline.

Value

A list containing projected data frames/matrices:

train_proj_train_with_pcs

Training data projected onto training PCs (data.frame)

test_proj_train_with_pcs

Test data projected onto training PCs (data.frame)

test_proj_clust_pcs

Test data projected onto clustering PCs (matrix)

train_proj_clust_pcs

Training data projected onto clustering PCs (data.frame); only present when compute_train_eval = TRUE

Details

Identifies features that are common between the training and test data sets. Extracts the PCA loadings from the training data for common features. Both training and test data are projected onto these loadings. Data is projected onto loadings for 2 PC sets (odd and even)