scSorter

Fork of cran/scSorter — adds Seurat v5 object support via RunScSorter().

scSorter is an R package that implements a semi-supervised algorithm for assigning cells to known cell types in single-cell RNA sequencing (scRNA-seq) data, based on user-provided marker genes.

The method is described in:

Guo, H., and Li, J. (2021). scSorter: assigning cells to known cell types according to known marker genes. Genome Biology, 22, 11.

What's New in This Fork

The original scSorter requires raw expression matrices. This fork adds native Seurat v5 support via RunScSorter(), so you can run the algorithm directly on a Seurat object without manually extracting and preprocessing the expression matrix.

Feature	Upstream	This Fork
Raw matrix input (scSorter())	✅	✅
Seurat v5 object input (RunScSorter())	❌	✅
Automatic variable feature selection	❌	✅
Sets Seurat identity class automatically	❌	✅
Expression filtering by detection rate (min_pct)	❌	✅
Allowing all markers to be missed in object for certain cell types	❌	✅

Installation

# install.packages("remotes")
remotes::install_github("pwwang/scSorter")

Quick Start — Seurat (Recommended)

library(Seurat)
library(scSorter)

# Your Seurat object with FindVariableFeatures already run
# anno: data frame with columns "Type" and "Marker"
result_obj <- RunScSorter(
  object = seurat_obj,
  anno = anno,
  top_vf = 2000,     # use top 2000 variable features
  min_pct = 0.1      # gene must be detected in ≥10% of cells
)

# Cell type predictions are stored in the metadata and set as the active identity
table(result_obj$scSorter_celltype)
DimPlot(result_obj, label = TRUE)

RunScSorter() Arguments

Argument	Description	Default
object	A Seurat object (must have FindVariableFeatures run)	—
anno	Marker gene annotation with columns Type, Marker, plus optional Weight	—
layer	Seurat layer for expression data (NULL = default layer)	NULL
assay	Seurat assay to use (NULL = default assay)	NULL
top_vf	Number of top variable features to include (NULL = all)	2000
min_pct	Minimum fraction of cells expressing a gene to retain it	0.1
set_ident	Set predicted cell types as active Seurat identity	TRUE
name	Metadata column name for predictions	"scSorter_celltype"
...	Additional arguments passed to scSorter()	—