Conversions: h5Seurat and AnnData

This vignette showcases how to convert from Seurat object to AnnData files via an intermediate step thorugh h5Seurat files. This allows interoperability between Seurat and Scanpy

library(Seurat)
library(SeuratData)
library(SeuratDisk)

Converting from Seurat to AnnData via h5Seurat

To showcase going from a Seurat object to an AnnData file, we'll use the processed version of the PBMC 3k dataset, available on SeuratData; this dataset was created following Seurat's PBMC 3k tutorial

InstallData("pbmc3k")
data("pbmc3k.final")
pbmc3k.final
#> An object of class Seurat 
#> 13714 features across 2638 samples within 1 assay 
#> Active assay: RNA (13714 features, 2000 variable features)
#>  2 dimensional reductions calculated: pca, umap

To see how this dataset was generated, please run ?pbmc3k.final

Converting the Seurat object to an AnnData file is a two-step process. First, we save the Seurat object as an h5Seurat file. For more details about saving Seurat objects to h5Seurat files, please see this vignette; after the file is saved, we can convert it to an AnnData file for use in Scanpy. Full details about the conversion processes are listed in the manual page for the Convert function

SaveH5Seurat(pbmc3k.final, filename = "pbmc3k.h5Seurat")
Convert("pbmc3k.h5Seurat", dest = "h5ad")

We can view the AnnData file in Scanpy by using the read_h5ad function

import scanpy
#> /home/paul/.local/lib/python3.6/site-packages/anndata/_core/anndata.py:21: FutureWarning: pandas.core.index is deprecated and will be removed in a future version.  The public classes are available in the top-level namespace.
#>   from pandas.core.index import RangeIndex
#> /home/paul/.local/lib/python3.6/site-packages/numba/core/errors.py:144: UserWarning: Insufficiently recent colorama version found. Numba requires colorama >= 0.3.9
#>   warnings.warn(msg)
adata = scanpy.read_h5ad("pbmc3k.h5ad")
adata
#> AnnData object with n_obs × n_vars = 2638 × 13714 
#>     obs: 'orig.ident', 'nCount_RNA', 'nFeature_RNA', 'seurat_annotations', 'percent.mt', 'RNA_snn_res.0.5', 'seurat_clusters'
#>     var: 'vst.mean', 'vst.variance', 'vst.variance.expected', 'vst.variance.standardized', 'vst.variable'
#>     uns: 'neighbors'
#>     obsm: 'X_pca', 'X_umap'
#>     varm: 'PCs'

Converting from AnnData to Seurat via h5Seurat

To shocwcase going from an AnnData file to a Seurat object, we'll use a processed version of the PBMC 3k dataset; this dataset was processed using Scanpy following Scanpy's PBMC 3k tutorial

url <- "https://seurat.nygenome.org/pbmc3k_final.h5ad"
curl::curl_download(url, basename(url))

To see how this dataset was created, please see this script

Converting the AnnData file to a Seurat object is a two-step process. First, convert the AnnData file to an h5Seurat file using the Convert function; full details about the conversion process are listed in the manual page. Then, we load the h5Seurat file into a Seurat object; for more details about loading Seurat objects from h5Seurat files, please see this vignette

Convert("pbmc3k_final.h5ad", dest = "h5seurat", overwrite = TRUE)
pbmc3k <- LoadH5Seurat("pbmc3k_final.h5seurat")
pbmc3k
#> An object of class Seurat 
#> 13714 features across 2638 samples within 1 assay 
#> Active assay: RNA (13714 features, 0 variable features)
#>  2 dimensional reductions calculated: pca, umap