Skip to contents

Reduce the size of a Seurat object

Source: R/Right_DietSeurat.R
Right_DietSeurat.Rd

As of Seurat v5 release, DietSeurat does not remove data and scale.data layers, resulting in objects with little to no reduction in size. This function was created with the purpose to restore DietSeurat's proper behavior until the function is fixed by Seurat's dev team, as well as to offer a few new options such as the ability to subset cells or keep variable features.

Usage

Right_DietSeurat(
  seurat_object,
  idents = NULL,
  cells = NULL,
  features = NULL,
  dimreducs = NULL,
  graphs = NULL,
  variable.features = FALSE,
  misc = TRUE,
  split.counts.layer = NULL,
  data.layer = FALSE,
  scale.layer = FALSE,
  SCTAssay = FALSE
)

Arguments

seurat_object

A Seurat object.

idents

Character. The names of one or several metadata (for example, 'orig.ident', 'seurat_clusters', etc) to keep in the diet object. If NULL, all metadata will be kept.

cells

Character. The names of one or several cells to keep in the diet object. If NULL, all cells will be kept.

features

Character. The names of one or several features to keep in the diet object. If NULL, all features will be kept.

dimreducs

Character. The names of one or several reductions to keep in the diet object. If NULL, all reductions will be removed. Please note that if you subset features and remove all from which principal components (PC) were calculated, the 'PCA' reduction will not be kept, even if it is provided, as the feature.loadings slot will be empty. This is intended behavior.

graphs

Character. The names of one or several graphs to keep in the diet object. If NULL, all graphs will be removed.

variable.features

Logical. If TRUE, the variable features slot will be kept in the diet object. Please note that if you keep only a subset of features, the variable features will also be subset, and if you remove all features from which variable features were calculated, the variable features will not be kept even if TRUE. This is intended behavior.

misc

Logical. If TRUE, the misc slot will be kept in the diet object.

split.counts.layer

Character. The name of a metadata (for example, 'orig.ident', 'seurat_clusters', etc) to split the 'counts' layer in the 'RNA' assay by if you need to in your downstream analysis. If NULL, the diet object will have a single 'counts' layer, even if the original object had split 'counts' layers.

data.layer

Logical. If TRUE, the 'data' layer in the 'RNA' assay will be kept in the diet object if it is present. As with the 'counts' layer, if there are split 'data' layers, they will be joined into a single 'data' layer unless split.counts.layer is provided.

scale.layer

Logical. If TRUE, the 'scale.data' layer in the 'RNA' assay will be kept in the diet object if it is present.

SCTAssay

Logical. If TRUE, the 'SCT' assay will be kept in the diet object if it is present.

Value

A Seurat object, hopefully smaller, with class Assay5 'RNA' assay and specified layers and slots.