Seurat define manually a cluster and find markers

Question

I am aware of this question Manually define clusters in Seurat and determine marker genes that is similar but I couldn't make tit work for my use case.

So I have a single cell experiments and the clustering id not great I have a small groups of 6 cells (I know it is extremely small, but nonetheless I would like to make the most of it) that are clearly isolate in UMAP and display marker that I am interresting in.

I managed to get those cells names in barcode (e.g.: AGCCAGCTCGCTTTAT).

I would like to compute differential expression of those cells against all others groups (using FindMarkers or FindAllMarkers).

I tried to manually define the cluster in: [email protected]$seurat_clusters But using FindMarkers (it says this groups don't exist)

then I try to modify [email protected]

but then it could not found metadata attached to my newly define group.

I am kind of stuck and any help is warmly welcome

Code

My R is quite rusty...

names_ # a vector with the names of the cell I want as a cluster

# my seurat object have 5 cluster 0..4. So I want to add a 6th clusters with id 5

#First attempt manually modify the vector active.ident and cluster
# they are both factor so we need to add a factor level
levels([email protected]) = c(levels([email protected]), 5)
levels([email protected]$seurat_clusters) = c(levels([email protected]$seurat_clusters), 5)
# then modify them :
[email protected][which(row.names([email protected]) %in%  names_)] = 5
[email protected]$seurat_clusters[which(row.names([email protected]) %in%  names_)] = 5

FindMarkers(s_19270BC, ident.1 = 5)
 

So This morning it works... I don't know what I have done yesterday...

EDIT: I added the Code I used

Answer 1

It would have been better if you have shared your code as well as the error messages you got format these as such.

As far as I can understand, your problem is that you failed to make your cluster as default ids. For this you can use:

seurat_obj <- SetIdent(seurat_obj, [email protected]$column_of_interest).

The cluster identities are stored in the @meta.data slot of your object.