Skip to contents

Prepare a Seurat Object with Selected Genes and PCA

prepare.seurat.Rd

This function scales the data and performs Principal Component Analysis (PCA) on a Seurat object using a specified set of genes. A placeholder for PC regression settings for co-expression inference is created and stored in the Seurat object's misc slot under NNet.setting.

Usage

prepare.seurat(
  seurat.obj,
  genes,
  npcs = 100,
  truncated = TRUE,
  ScaleData.ctrl = list(),
  RunPCA.ctrl = list()
)

Arguments

seurat.obj

A Seurat object containing a data layer.

genes

A character vector of gene names to use for scaling and PCA. Only genes present in seurat.obj will be used.

npcs

An integer specifying the maximum number of principal components to compute. Default is 100.

truncated

A logical value indicating whether to select a subset of significant PCs based on their standard deviation. If TRUE, only significant PCs are used. Default is TRUE.

ScaleData.ctrl

A list of additional parameters to pass to ScaleData. The features and verbose arguments are set automatically.

RunPCA.ctrl

A list of additional parameters to pass to RunPCA. The features, npcs, and verbose arguments are set automatically.

Value

A Seurat object with PC regression settings stored in its misc slot as a list named NNet.setting. The list includes:

pcs

A matrix of PC embeddings for all cells. Rows correspond to cells, and columns correspond to PCs.

loadings

A matrix of PC loadings. Rows correspond to genes, and columns correspond to PCs.

p

Placeholder for the KNN affinity matrix.

nn.idx

Placeholder for the indices of nearest neighbors.

nn.w

Placeholder for the weights of nearest neighbors.

all.cells

A character vector of cells used to ran PCA.

all.genes

A character vector of genes used to ran PCA.

cells

Placeholder for selected cells.

predictors

Placeholder for selected predictor genes.

responses

Placeholder for selected response genes.

genes

Placeholder for the set of selected genes.

lra

Placeholder for the low-rank approximation matrix.

scale.gene

Placeholder for global gene scales.

nn.scale.gene

Placeholder for gene-level variance scaling.

nn.scale.pc

Placeholder for PC-level variance scaling.

n.eff

Placeholder for effective neighborhood sizes.

Details

This function prepares a Seurat object for PC regression analysis. It filters the provided gene list to include only those present in the Seurat object, scales the data using ScaleData, and performs PCA using RunPCA. If truncated = TRUE, the number of significant PCs is determined using their standard deviations. Users can customize the scaling and PCA processes by providing additional control parameters via ScaleData.ctrl and RunPCA.ctrl.

The prepared Seurat object is stored with placeholders for regression settings to enable downstream co-expression analysis using NNet.

See also

ScaleData, RunPCA

Examples

# Select genes for PC regression
genes <- select.gene(seurat.obj)$genes
#> Error: object 'seurat.obj' not found

# Prepare the Seurat object with default settings
seurat.obj <- prepare.seurat(seurat.obj, genes, npcs = 30)
#> Error: object 'genes' not found

# Customize scaling and PCA parameters
seurat.obj <- prepare.seurat(seurat.obj, genes, npcs = 50,
                             ScaleData.ctrl = list(do.center = FALSE))
#> Error: object 'genes' not found