Hello, I noticed in the tutorial for GIMVI there is an option for specifying the labels in each dataset:
I’m wondering whether this is supposed to be the cluster labels, as in each dataset was clustered prior and for some reason that clustering info may be useful in the imputation process. or whether these cluster labels should be putative annotations that roughly match each other, like whether I could identify similar cell types across the datasets and they should be labeled the same name.
I didn’t see anything about cluster labels in the paper, so ultimately I’m a bit confused on how this is factored into the VAE / bayesian probability framework. Any thoughts would be appreciated!