Basic Usage

Introduction

This vignette demonstrates the basic usage of recall for calibrated clustering in single-cell RNA-sequencing analysis. The package is compatible with both Seurat V4 and V5, and works on all major platforms (Linux, macOS, Windows).

Setup

library(Seurat)
library(recall)

Loading Data

You can load your own single-cell data or use publicly available datasets:

# Option 1: Load from a 10X Genomics directory
# seurat_obj <- Read10X(data.dir = "path/to/filtered_feature_bc_matrix/")
# seurat_obj <- CreateSeuratObject(counts = seurat_obj)

# Option 2: Load from an RDS file
# seurat_obj <- readRDS("your_seurat_object.rds")

# For this example, we assume you have a Seurat object named 'seurat_obj'

Standard Preprocessing

Before running recall, perform standard Seurat preprocessing:

seurat_obj <- NormalizeData(seurat_obj)
seurat_obj <- FindVariableFeatures(seurat_obj, nfeatures = 2000)
seurat_obj <- ScaleData(seurat_obj)
seurat_obj <- RunPCA(seurat_obj)
seurat_obj <- FindNeighbors(seurat_obj, dims = 1:10)
seurat_obj <- RunUMAP(seurat_obj, dims = 1:10)

Running recall

The recall algorithm can be run with a single function call as a drop-in replacement for the Seurat function FindClusters:

seurat_obj <- FindClustersRecall(seurat_obj, resolution_start = 0.8)

Accessing Results

The recall clusters are set as the default identities of the returned Seurat object:

# View cluster distribution
table(Idents(seurat_obj))

# Cluster labels are also stored in metadata
head(seurat_obj@meta.data$recall_clusters)

Visualization

# UMAP visualization with cluster labels
DimPlot(seurat_obj, label = TRUE)

# Or explicitly specify the recall clusters
DimPlot(seurat_obj, group.by = "recall_clusters", label = TRUE)

Comparison with Standard Clustering

Compare recall results with standard Seurat clustering:

# Standard Seurat clustering
seurat_standard <- FindClusters(seurat_obj, resolution = 0.8)

# Compare number of clusters
cat("Standard clustering:", length(unique(Idents(seurat_standard))), "clusters\n")
cat("recall clustering:", length(unique(seurat_obj$recall_clusters)), "clusters\n")

Next Steps

For more advanced usage, see:

Algorithm & Methodology: Understanding the knockoff filter approach
Visualization Guide: Creating publication-quality figures
Advanced Usage: Parameter tuning and alternative methods

Session Info

sessionInfo()

Zaoqu Liu

2026-01-24