6  ScType (Automatic annotation)

Cluster annotation-subsetted for cells > 2000 features

# devtools::install_github("immunogenomics/presto")

suppressPackageStartupMessages({
library(tidyverse)
library(Seurat)
library(patchwork)
library(presto)
library(kableExtra)
    })

ScType is a computational method for automated selection of marker genes based merely on scRNA-seq data (Ianevski, Giri, and Aittokallio 2022). ScType combines a large, curated marker gene database with a cell-type specificity scoring system to accurately assign cell identities. The ScType marker database integrates information from CellMarker, PanglaoDB, and manual curation to include 3,980 markers for 194 human cell types and 4,212 markers for 194 mouse cell types, along with both positive and negative markers across 17 tissues per species. The specificity score quantifies how uniquely a gene identifies a cell type within a tissue by considering its presence across multiple cell types, with higher scores reflecting greater specificity. This score is used to weight gene expression during annotation, enhancing the reliability of cell-type assignments.

alldata <- readRDS("data/20250620-seurat_integrated_UMAP_nFeat>600.Rds")

library(HGNChelper)
library(openxlsx)
source("https://raw.githubusercontent.com/IanevskiAleksandr/sc-type/master/R/sctype_wrapper.R")

alldata <- run_sctype(alldata,
                      assay = "SCT",
                      scaled = F,
                      known_tissue_type="Brain",
                      custom_marker_file="https://raw.githubusercontent.com/IanevskiAleksandr/sc-type/master/ScTypeDB_full.xlsx",
                      name="sctype_classification")

save(alldata, file = "data/20250620-seurat_integrated_UMAP_nFeat-ScTypeannotated.Rdata")

load("data/20250620-seurat_integrated_UMAP_nFeat-ScTypeannotated.Rdata")

alldata$ribo_percent <- PercentageFeatureSet(alldata,assay = "SCT", pattern = "^Rp[ls]")

6.1 UMAP

Idents(alldata) <- alldata@meta.data$sctype_classification
DimPlot(alldata, label = TRUE, repel = T) + 
    NoLegend()

FeaturePlot(alldata, features = c("ribo_percent","nFeature_SCT","Malat1"),
            split.by = "type")

DimPlot(alldata, label = TRUE, repel = T, split.by = "orig.ident",
        alpha = 0.5, ncol = 3) +
    NoLegend()

DimPlot(alldata, label = TRUE, repel = T, split.by = "type",
        alpha = 0.5) +
    NoLegend()

6.2 Cell markers

6.2.1 Microglia

VlnPlot(alldata,features = "Cx3cr1", alpha = 0.01, y.max = 7,
        split.by = "orig.ident")

VlnPlot(alldata,features = "Tmem119",alpha = 0.01, y.max = 5, 
        split.by = "orig.ident")

VlnPlot(alldata,features = "Lpl",alpha = 0.01, y.max = 5, 
        split.by = "orig.ident")

VlnPlot(alldata,features = "Aif1",alpha = 0.01, y.max = 5, 
        split.by = "orig.ident")

VlnPlot(alldata,features = "ribo_percent",alpha = 0.01, 
        split.by = "orig.ident")

6.2.2 Astrocytes

VlnPlot(alldata,features = "Aldh1l1", alpha = 0.03, y.max = 8, 
        split.by = "orig.ident")

6.2.3 Oligodendrocyte

VlnPlot(alldata,features = "Olig1", alpha = 0.03, 
        split.by = "orig.ident")

VlnPlot(alldata,features = "Mog", alpha = 0.03, 
        split.by = "orig.ident")

Ianevski, Aleksandr, Anil K. Giri, and Tero Aittokallio. 2022. “Fully-Automated and Ultra-Fast Cell-Type Identification Using Specific Marker Combinations from Single-Cell Transcriptomic Data.” Nature Communications 13 (1). https://doi.org/10.1038/s41467-022-28803-w.