I have a scRNA-seq that I want to annotate using the Linnarson's brain cell atlas. To that end, I am using the Label Transfer pipeline from Seurat, which you can find here: https://satijalab.org/seurat/v3.0/pancreas_integration_label_transfer.html . This is my code:

# Connect to loom object
brain_loom <- connect(filename = "data/TM_brain_ref.loom")
brain_loom 

# Convert to Seurat
brain_seu <- as.Seurat(brain_loom)

# Pre-processing
brain_seu <- NormalizeData(
  object = brain_seu, 
  normalization.method = "LogNormalize",
  scale.factor = 10000, 
  display.progress = TRUE
)
brain_seu <- FindVariableFeatures(brain_seu, selection.method = "vst", nfeatures = 2000)
brain_seu <- ScaleData(brain_seu)
brain_seu <- RunPCA(brain_seu, features = VariableFeatures(object = brain_seu))
saveRDS(brain_seu, "results/R_objects/brain_reference_Seurat.rds")

# Project data
anchors <- FindTransferAnchors(reference = brain_seu, query = epid_seu, dims = 1:30)

However, when executing the FindTransferAnchors() function, I get the following error:

Performing PCA on the provided reference using 2000 features as input.
Projecting PCA
Error in FastSparseRowScaleWithKnownStats(mat = proj.data, mu = feature.mean,  : 
  Not an S4 object.

Do you have any idea on what is the underlying problem?

Thanks a lot!