Interoperability on Seurat V5

Seurat is undoubtedly the most used package for single-cell analysis. However, the single-cell field has changed towards new programming languages and packages (e.g., Scanpy), which leads to an issue regarding interoperability , i.e., the ability of software to exchange and make use of information. Here, we show how to quickly convert Seurat objects from ongoing projects to AnnData (Scanpy data object).

1. Loading requirements

library(Seurat)
library(dplyr)
library(sceasy)

# Setting Assay version 5
options(Seurat.object.assay.version = "v5")

# Load the PBMC dataset - available at 10x datasets website
pbmc.data <- Read10X(
    data.dir = "/Users/affaustino/Projects/UTILS/quarto_blog/data/filtered_gene_bc_matrices/hg19")

# Initialize the Seurat object with the raw (non-normalized data).
pbmc_seurat <- CreateSeuratObject(
  counts = pbmc.data, project = "pbmc3k", min.cells = 3, min.features = 200)

# The [[ operator can add columns to object metadata. This is a great place to stash QC stats
pbmc_seurat[["percent_mt"]] <- PercentageFeatureSet(pbmc_seurat, pattern = "^MT-")

pbmc_seurat <- pbmc_seurat %>%
  NormalizeData(normalization.method = "LogNormalize", scale.factor = 10000) %>%
  FindVariableFeatures(selection.method = "vst", nfeatures = 2000) %>%
  ScaleData()

2. Inspecting object

Once we have the Seurat object, we can quickly inspect its structure. Why should we do it? For education, it is nice to understand what slots will be kept over the conversion process.

pbmc_seurat

An object of class Seurat 
13714 features across 2700 samples within 1 assay 
Active assay: RNA (13714 features, 2000 variable features)
 3 layers present: counts, data, scale.data

pbmc_seurat@assays

$RNA
Assay (v5) data with 13714 features for 2700 cells
Top 10 variable features:
 ISG15, CPSF3L, MRPL20, ATAD3C, C1orf86, RER1, RP3-395M20.9, LRRC47,
GPR153, TNFRSF25 
Layers:
 counts, data, scale.data

3. Converting to SingleCellExperiment

In our experience, the SingleCellExperiment object works as an intermediate layer between Seurat and AnnData. In short, it will ensure that most of the data is maintained among formats.

pbmc_sce <- as.SingleCellExperiment(
    pbmc_seurat)

4. Saving as AnnData

Finally, we will use sceasy for converting SingleCellExperiment to Anndata. sceasy is a package that helps easily convert different single-cell data formats to each other. Converting to AnnData creates a file that can be directly used in cellxgene which is an interactive explorer for single-cell transcriptomics datasets.

convertFormat(pbmc_sce, from="sce", to="anndata",
                       outFile='/Users/affaustino/Projects/UTILS/quarto_blog/data/pbmc.h5ad')

AnnData object with n_obs × n_vars = 2700 × 13714
    obs: 'nCount_RNA', 'nFeature_RNA', 'percent_mt'
    var: 'name'

The figures are derived from 1, and 2