Apply Chi-squared normalization to a Seurat object

This function applies Chi-squared normalization to a (sample-level) Seurat object. It calculates the expected density (count of cells) for each landmark-sample pair assuming random distribution, then normalizes observed counts using: (observed - expected) / sqrt(expected). The resulting values show how much each landmark's density deviates from the expected baseline density values. Positive values indicate higher-than-expected density, negative values indicate lower-than-expected density.

Usage

NormalizeChiSquared(object, ...)

# Default S3 method
NormalizeChiSquared(object = NULL, verbose = TRUE, ...)

# S3 method for class 'Assay'
NormalizeChiSquared(object, verbose = TRUE, ...)

# S3 method for class 'StdAssay'
NormalizeChiSquared(
  object,
  layer = "counts",
  save = "data",
  verbose = TRUE,
  ...
)

# S3 method for class 'Seurat'
NormalizeChiSquared(object, assay = NULL, verbose = TRUE, ...)

Arguments

object

a count matrix (a landmark-sample density matrix), Assay, StdAssay, or Seurat object. Values must be raw counts (non-negative whole numbers)

...

Additional arguments

verbose

display progress bar for normalization procedure

layer

Layer to use (for StdAssay objects, and passed on by the Seurat method). It must resolve to a single layer whose name starts with "counts"; for split layers, run JoinLayers first or name one layer

save

Layer to save results to (for StdAssay objects, and passed on by the Seurat method). Its name cannot start with "counts"

assay

Name of assay to use (for Seurat objects)

Value

Normalized object of the same class as input

Examples

if (FALSE) { # \dontrun{
# Normalize a sample-level Seurat object
sample_obj <- NormalizeChiSquared(sample_obj)
} # }