vignettes/convert-anndata.Rmd

---
title: 'Conversions: h5Seurat and AnnData'
---

```{r setup, include = FALSE}
python3 <- Sys.which(names = "python3")
library(reticulate)
use_python(python = python3)
knitr::opts_chunk$set(
  message = FALSE,
  warning = FALSE,
  collapse = TRUE,
  comment = "#>"
)
# options(SeuratData.repo.use = "https://seurat.nygenome.org")
```

```{r cleanup, echo=FALSE, results='hide'}
outs <- c(
  "pbmc3k.h5Seurat",
  "pbmc3k.h5ad",
  "pbmc3k_final.h5ad",
  "pbmc3k_final.h5seurat"
)
file.remove(outs)
```

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](https://scanpy.readthedocs.io/)

```{r packages}
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](https://github.com/satijalab/seurat-data); this dataset was created following [Seurat's PBMC 3k tutorial](https://satijalab.org/seurat/v3.1/pbmc3k_tutorial.html)

```{r get_data}
InstallData("pbmc3k")
data("pbmc3k.final")
pbmc3k.final
```

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](https://mojaveazure.github.io/seurat-disk/reference/SaveH5Seurat.html) as an h5Seurat file. For more details about saving `Seurat` objects to h5Seurat files, please see [this vignette](https://mojaveazure.github.io/seurat-disk/articles/h5Seurat-load.html); after the file is saved, we can [convert](https://mojaveazure.github.io/seurat-disk/reference/Convert.html) it to an AnnData file for use in Scanpy. Full details about the conversion processes are listed in the [manual page](https://mojaveazure.github.io/seurat-disk/reference/Convert.html) for the `Convert` function

```{r convert_seurat}
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

```{python load_adata}
import scanpy
adata = scanpy.read_h5ad("pbmc3k.h5ad")
adata
```

## 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](https://scanpy-tutorials.readthedocs.io/en/latest/pbmc3k.html)

```{r get_adata}
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](https://gist.github.com/mojaveazure/922aa904b9ac212e627f522c2b816a52)


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](https://mojaveazure.github.io/seurat-disk/reference/Convert.html); full details about the conversion process are listed in the [manual page](https://mojaveazure.github.io/seurat-disk/reference/Convert.html). Then, we [load the h5Seurat file](https://mojaveazure.github.io/seurat-disk/reference/LoadH5Seurat.html) into a `Seurat` object; for more details about loading `Seurat` objects from h5Seurat files, please see [this vignette](https://mojaveazure.github.io/seurat-disk/articles/h5Seurat-load.html)

```{r convert_adata}
Convert("pbmc3k_final.h5ad", dest = "h5seurat", overwrite = TRUE)
pbmc3k <- LoadH5Seurat("pbmc3k_final.h5seurat")
pbmc3k
```