ScaleData() fails with “all cells in assays must be present in the Seurat object” error

[BenjaminDEMAILLE]

Title: ScaleData() fails with “all cells in assays must be present in the Seurat object” error

Description:
Calling ScaleData() on a subsetted Seurat object leads to an error about assay cell mismatch, even though the object was previously subsetted and appears valid.

Code to reproduce:

NeuroProx_sub <- ScaleData(NeuroProx_sub, assay = "RNA", features = markers_f$gene)

Output:

Centering and scaling data matrix
  |==========================================================================================| 100%
Error in validObject(object = x) : 
  invalid class “Seurat” object: all cells in assays must be present in the Seurat object
Expected behavior:
ScaleData() should complete without error on a valid, subsetted Seurat object if the features and assay are correctly specified.

Troubleshooting attempted:

Verified that markers_f$gene contains genes present in the RNA assay.
Checked that NeuroProx_sub is a valid Seurat object using ValidObject().
Assay "RNA" exists and is the default assay in NeuroProx_sub.
Environment:

Seurat version: 5.3
SeuratObject version: 5.1
R version: 4.5.0

Suggestion:
The error message suggests a mismatch between the cells in the RNA assay and the Seurat object. This might be due to residual cells in the assay slot after subsetting. If so, ScaleData() should better handle such cases or validate the object earlier with a more informative message.

[samuel-marsh]

Hi @BenjaminDEMAILLE,

Can you provide reproducible example (ideally using public object (eg from SeuratData package)? I cannot replicate issues currently in my testing suggesting it may be issue with the Seurat object you are using.

Best,
Sam

[BenjaminDEMAILLE]

It don’t happen with seurat 5.2

[BenjaminDEMAILLE]

Hi @BenjaminDEMAILLE,

Can you provide reproducible example (ideally using public object (eg from SeuratData package)? I cannot replicate issues currently in my testing suggesting it may be issue with the Seurat object you are using.

Best, Sam

The full code is

NeuroProx_sub <- subset(NeuroProx, downsample = 1000)
markers_f
NeuroProx_sub <- ScaleData(NeuroProx_sub, assay = "RNA", features = markers_f$gene)
Centering and scaling data matrix
  |===========================================================================================================| 100%
Error in validObject(object = x) : 
  invalid class “Seurat” object: all cells in assays must be present in the Seurat object

[BenjaminDEMAILLE]

The problem is due to the subset :

> NeuroProx <- NeuroProx[, NeuroProx$Phase == "S"]
> validObject(NeuroProx)
Error in validObject(NeuroProx) : 
  invalid class “Seurat” object: all cells in assays must be present in the Seurat object
> NeuroProx <- qread("20250401_NeuroProx_velo.qs", nthreads = detectCores())
> validObject(NeuroProx)
[1] TRUE

[samuel-marsh]

Ok thanks. I’ll look into it.