Label transfer without any batch removal in the reference

[tvegawaichman]

Hello!
I'm trying to use scArches to do a label transfer between a reference and a query, and I have a few questions:

Is this normalization enough, or should it be done in a specific way?

sc.pp.normalize_total(adata, target_sum=1e4)
# Logarithmize the data:
sc.pp.log1p(adata)

Then, I want to use scPoli, but the model is asking about the condition_key. Is it enough to create a dummy variable where the condition for all the labels in the reference is the same? (All my cells come from the same condition-batch). Like:
adata.obs['condition'] = 'none'
What does cell_type_key refer to? Should I specify the cell-type column that I want to use to transfer the labels here?

I see that you recommend scANVI also for annotation. Could any other method in scArches (like scGen) be useful for this task?

Thanks!