seurat-v5

Seurat v5: Multi-Modal Single-Cell Analysis

Overview

Seurat v5 is the latest version of the Seurat package, introducing significant improvements for multi-modal single-cell data analysis. It provides a unified framework for analyzing and integrating multiple data types (gene expression, protein, chromatin accessibility) from the same cells.

When to Use This Skill

This skill should be used when:

• Analyzing CITE-seq or REAP-seq data (RNA + protein)
• Working with multiple modalities from the same cells
• Performing multi-omics integration
• Using the latest Seurat methods
• Working in R-based workflows

Quick Start

Installation

# Install Seurat v5
install.packages("Seurat")
install.packages("SeuratData")

# Verify version
packageVersion("Seurat")  # Should be 5.x.x

Basic Multi-Modal Setup

library(Seurat)

# Create object with RNA and protein data
cbmc <- CreateSeuratObject(counts = cbmc.rna, assay = "RNA")
cbmc[["ADT"]] <- CreateAssayObject(counts = cbmc.adt)

# Check assays
cbmc
# An object of class Seurat
# 34129 features across 8617 samples within 2 assays
# Active assay: RNA (23690 features, 0 variable features)
#  2 other assays present: ADT

Working with Multi-Modal Data

RNA Analysis

# Switch to RNA assay
DefaultAssay(cbmc) <- "RNA"

# Standard RNA analysis
cbmc <- NormalizeData(cbmc)
cbmc <- FindVariableFeatures(cbmc, nfeatures = 2000)
cbmc <- ScaleData(cbmc)
cbmc <- RunPCA(cbmc, npcs = 30)
cbmc <- RunUMAP(cbmc, dims = 1:30)

Protein (ADT) Analysis

# Switch to ADT assay
DefaultAssay(cbmc) <- "ADT"

# ADT-specific processing
cbmc <- NormalizeData(cbmc, normalization.method = "CLR", margin = 2)
cbmc <- ScaleData(cbmc)

# Run PCA on protein data
cbmc <- RunPCA(cbmc, reduction.name = "adtpca", dims = 1:10)

# Find neighbors using protein
cbmc <- FindMultiModalNeighbors(
  cbmc,
  reduction.list = list("pca", "adtpca"),
  dims.list = list(1:30, 1:10),
  modality.weight.name = "RNA.weight"
)

# UMAP using both modalities
cbmc <- RunUMAP(cbmc, nn.name = "weighted.nn", reduction.name = "wnn.umap")

# Plot
DimPlot(cbmc, reduction = "wnn.umap", group.by = "celltype")