Maximum Posterior Probability

[xuebaliang]

Hello, I have a question about the cell type classes setting. In your paper, you say "By default scANVI classifier sets the number of classes to the same number of cell types in the merged dataset. In the case of N observed labels from the annotated dataset and one unannotated dataset (thus the cell type label is “Unlabeled”) scANVI assumes N + 1 classes. ", I am very confused with these two sentences. In the first sentences, I think you say the classes number is the sum of annotated dataset and unannotated dataset. But the second sentence means that the classes number is the annotated labels plus one, but if the unannotated dataset has more than one cell type not in annotated dataset? Are they all belong to "unknown" cell type?? I want to know the output form of scanvi in cell type annotation. Thanks!