Skip to contents

Documentation | Preprint | Discord

MuDataSeurat is a package that provides I/O funcitonality for .h5mu files and Seurat objects.

You can learn more about multimodal data containers in the mudata documentation.

Installation

remotes::install_github("pmbio/MuDataSeurat")

Quick start

MuDataSeurat provides a set of I/O operations for multimodal data.

MuDataSeurat implements WriteH5MU() that saves Seurat objects to .h5mu files that can be further integrated into workflows in multiple programming languages, including the muon Python library and the Muon.jl Julia library. ReadH5MU() reads .h5mu files into Seurat objects.

MuDataSeurat currently works for Seurat objects of v3 and above.

Writing files

Start with an existing dataset, e.g. a Seurat object with CITE-seq data:

library(SeuratData)
InstallData("bmcite")
bm <- LoadData(ds = "bmcite")

WriteH5MU() allows to save the object into a .h5mu file:

library(MuDataSeurat)
WriteH5MU(bm, "bmcite.h5mu")

Please note that only standardised parts of the object are written to the file, and extra information from specific methods, stored in the Seurat object, might be omitted upon writing the file.

Reading files

bm <- ReadH5MU("bmcite.h5mu")

Please note that only the intersection of cells is currently loaded into the Seurat object due to the object structure limitation. Multimodal embeddings (global .obsm slot) are loaded with the assay.used field set to the default assay. Embeddings names are changed in order to comply with R & Seurat requirements and conventions.

Relevant projects

Other R packages for multimodal I/O include: