In Bioconductor/ZarrExperiment: Manipulate Zarr archives as R objects

knitr::opts_chunk$set(
    collapse = TRUE,
    cache = TRUE
)

Introduction

The ZarrExperiment package is currently under construction. The purpose of the package is to build an interface between zarr (https://zarr.readthedocs.io/en/stable/) and the Bioconductor SummarizedExperiment family.

Installation

Install the package with

R -e "BiocManager::install('Bioconductor/ZarrExperiment')"

The ZarrExperiment package uses the zarr module from python. Configuring python currently requires a git clone.

git clone https://github.com/Bioconductor/ZarrExperiment

We suggest using a python virtual environment to configure python. There are different ways of establishing virtual environments.

venv

virtualenv

virtualenv is a distinct program. Do these from the command line.

1) Install python3 and virtualenv

2) Create the virtual environment.

virtualenv -p python3 ~/.virtualenvs/Bioconductor

3) Activate this virtual environment.

source ~/.virtualenvs/Bioconductor/bin/activate

4) Install the required python packages from the file python-requirements.txt in the package's base directory using pip.

pip install -r python-requirements.txt

When done, deactivate the virtual environment.

deactivate

6) Set the enviroment RETICULATE_PYTHON variable to use the Bioconductor virtual environment.

export RETICULATE_PYTHON=~/.virtualenvs/Bioconductor/bin/python

7) Test that zarr can be imported with reticulate.

R -e "reticulate::import('zarr')"

basilisk

Use

Load libraries used in this vignette

library(SummarizedExperiment)
library(tibble)
library(ZarrExperiment)

Point to a sample archive. Archives are folders with a collection of files.

fl <- system.file(
    package="ZarrExperiment", "extdata",
    "stahl-2016-science-olfactory-bulb.matrix.zarr"
)
dir(fl, recursive = TRUE, all.files = TRUE)

Load the archive and view, via the show() or tree() method, the available groups (datasets) in the archive. These groups are analogous to hdf5 datasets.

arr <- ZarrArchive(fl)
arr

The archive allows $ subsetting, including tab completion. Access the matrix group member and coerce it to an R (dense) matrix.

m <- t(as(arr$matrix, "matrix"))
m[1:5, 1:5]

The components of this archive (archive format is completely general, so it is not possible to write a 'smart' import function) can be accessed...

rowData <- tibble(
    gene_name = as(arr$gene_name, "matrix")
)
rowData

colData <- tibble(
    region_id = as(arr$region_id, "matrix"),
    x_region = as(arr$x_region, "matrix"),
    y_region = as(arr$y_region, "matrix")
)
colData

Form a SummarizedExperiment from these components:

se <- SummarizedExperiment(
    assays = list(count = m),
    rowData = rowData,
    colData = colData
)
se

The SummarizedExperiment object can then be used in standard R / Bioconductor single-cell and other work flows.

Acknowledgements

sessionInfo()

Bioconductor/ZarrExperiment documentation built on Oct. 9, 2024, 1:51 p.m.