Load a MuData H5MU file as a Seurat object

Read multimodal MuData (.h5mu) files and convert to a Seurat object with multiple assays. Uses native HDF5 reading with no external dependencies (no MuDataSeurat or Python required).

Usage

readH5MU(
  file,
  modalities = NULL,
  assay.names = NULL,
  restore.spatial = TRUE,
  verbose = TRUE
)

Arguments

file

Path to .h5mu file

modalities

Character vector of modality names to load. If NULL (default), loads all modalities found in the file

assay.names

Named vector mapping modality names to desired Seurat assay names. If NULL, uses standard mapping (rna->RNA, prot->ADT, atac->ATAC, etc.)

restore.spatial

Logical; if TRUE, attempts to restore spatial data from the h5mu file structure (coordinates, images, scalefactors)

verbose

Show progress messages

Value

A Seurat object with multiple assays corresponding to each modality

Details

The h5mu format stores multimodal data where each modality is stored as a separate AnnData-like structure under /mod/{modality_name}. This function:

• Reads each modality's expression matrix (X) and layers

• Creates Seurat assays with appropriate names

• Preserves global cell metadata from /obs

• Reads per-modality embeddings (obsm) and graphs (obsp)

• Restores spatial data if present

Modality Mapping

By default, modality names are mapped to standard Seurat assay names:

• rna -> RNA

• prot -> ADT

• atac -> ATAC

• spatial -> Spatial

• Other names are preserved as-is

Examples

if (FALSE) { # \dontrun{
# Load all modalities from an h5mu file
seurat_obj <- readH5MU("multimodal_data.h5mu")

# Load specific modalities only
seurat_obj <- readH5MU("multimodal_data.h5mu", modalities = c("rna", "prot"))

# Custom assay name mapping
seurat_obj <- readH5MU(
  "data.h5mu",
  assay.names = c(rna = "RNA", prot = "Protein")
)
} # }