Odd cell neighbourhood preservation scores after manually creating SinglecellExperiment from Seurat V5 assay

[ondina-draia]

Hello,

I encounter a problem with latest version of the Seurat object (V5 Assays) which is also SCTransformed, when I try to convert in SingleCellExperiment format, I get the following error.

P30.N.sce <- as.SingleCellExperiment(obj$P30.N)

Error in `GetAssayData()`:
! GetAssayData doesn't work for multiple layers in v5 assay

I tried to find a workaround to this trouble by manually creating the SingleCellExperiment object manually:

# Convert to V3/4 assay
obj$P30.N <- NormalizeData(obj$P30.N, assay = "RNA")
obj$P30.N[["RNA3"]] <- as(object = obj$P30.N[["RNA"]], Class = "Assay")

# Raw data stored in RNA3 as counts
counts <- GetAssayData(obj$P30.N, assay = "RNA3", slot = "counts")

# Extract normalized counts
#normalized_counts <- GetAssayData(obj$P30.N, assay = "RNA3", slot = "data")

# MetaData
meta_data <- [email protected]

# Create SingleCellExperiment object
P30.N.sce <- SingleCellExperiment(
  assays = list(counts = counts, logcounts = normalized_counts),
  colData = meta_data
)

# Extract PCA data
pca_coords <- Embeddings(obj$P30.N, reduction = "pca")

# Add PCA coordinates to the SingleCellExperiment object
reducedDim(P30.N.sce, "PCA") <- pca_coords

# If thereés UMAP embeddings, extract them:
if ("umap" %in% names(obj$P30.N@reductions)) {
  umap_coords <- Embeddings(obj$P30.N, reduction = "umap")
  reducedDim(P30.N.sce, "UMAP") <- umap_coords
}

It seems to work until I check the cell neighborhood preservation scores:

I don't have this error or this kind of behaviour with the V3 assays.

What should I do? What is the cause of this strange behaviour?