Converting single-cell data structures between Bioconductor and Python
In zellkonverter: Conversion Between scRNA-seq Objects

require(knitr)
library(BiocStyle)
opts_chunk$set(error = FALSE, message = FALSE, warning = FALSE)

Overview

This package provides a lightweight interface between the Bioconductor SingleCellExperiment data structure and the Python AnnData-based single-cell analysis environment. The idea is to enable users and developers to easily move data between these frameworks to construct a multi-language analysis pipeline across R/Bioconductor and Python.

Reading and writing H5AD files

The readH5AD() function can be used to read a SingleCellExperiment from a H5AD file. This can be manipulated in the usual way as described in the r Biocpkg("SingleCellExperiment") documentation.

library(zellkonverter)

# Obtaining an example H5AD file.
example_h5ad <- system.file("extdata", "krumsiek11.h5ad",
                            package = "zellkonverter")
readH5AD(example_h5ad)

We can also write a SingleCellExperiment to a H5AD file with the writeH5AD() function. This is demonstrated below on the classic Zeisel mouse brain dataset from the r Biocpkg("scRNAseq") package. The resulting file can then be directly used in compatible Python-based analysis frameworks.

library(scRNAseq)

sce_zeisel <- ZeiselBrainData()
out_path <- tempfile(pattern = ".h5ad")
writeH5AD(sce_zeisel, file = out_path)

Converting between `SingleCellExperiment` and `AnnData` objects

Developers and power users who control their Python environments can directly convert between SingleCellExperiment and AnnData objects using the SCE2AnnData() and AnnData2SCE() utilities. These functions expect that r CRANpkg("reticulate") has already been loaded along with an appropriate version of the anndata package. We suggest using the r Biocpkg("basilisk") package to set up the Python environment before using these functions.

library(basilisk)
library(scRNAseq)

seger <- SegerstolpePancreasData()
roundtrip <- basiliskRun(fun = function(sce) {
     # Convert SCE to AnnData:
     adata <- SCE2AnnData(sce)

     # Maybe do some work in Python on 'adata':
     # BLAH BLAH BLAH

     # Convert back to an SCE:
     AnnData2SCE(adata)
}, env = zellkonverter:::anndata_env, sce = seger)

Package developers can guarantee that they are using the same versions of Python packages as r Biocpkg("zellkonverter") by using the .AnnDataDependencies variable to set up their Python environments.