Skip to contents

run_batch_correction: Performs batch correction within Seurat v5 framework

run_batch_correction.Rd

Performs batch correction using Seurat v5, with additional dimensionality reduction (UMAP), clustering, and cluster-based cell type annotation.

Usage

run_batch_correction(
  so_in,
  npcs,
  species,
  resolution_list,
  method_in,
  reduction_in = c(0.2, 0.4, 0.5, 0.8, 1),
  vars_to_regress = NULL,
  conda_env = ""
)

Arguments

so_in

A merged Seurat object containing all samples prior to batch correction

npcs

The number of principal components to use for dimensionality reduction and neighbor identification

species

"hg19", "hg38", or "mm10". Used to determine databases used for cell type annotation

resolution_list

A vector of resolutions ranging from 0 to 2.0 for clustering. Smaller resolutions produce fewer clusters (Default)

method_in

A character string to indicate which batch correction method to use

reduction_in

A character string for naming the PCA produced

vars_to_regress

A character vector for variables to regress out when running SCTransform normalization and batch correction

conda_env

A character string for indicating which conda environment contains the necessary packages

Value

A batch corrected Seurat object with updated UMAP projections and clusters

Details

Utilizes the Seurat v5 framework to streamline batch correction using one of the following options:

  • SCVI

  • LIGER

  • Canonical Correlation Analysis (CCA, or Seurat "integrate")

  • Reciprocal PCA (RPCA)

  • Harmony Follows up with re-running dimensionality reduction with PCA, neighbor finding, and UMAP projection. Also runs clustering with the slow local moving algorithm at a series of resolutions selected by the user. Final step conducts cell type annotation with SingleR using the average gene expression vector of each cluster