Automatically train a single cell annotation model, using scVI-Tools. Data should be in .h5ad, .h5, mtx, or .csv format. Uses the Anndata python package, which natively supports .h5ad meaning that file format is most suitable. The target should be the feature you want to label: most commonly this will be cell type. Please ensure that the column with these values is selected. A model will be trained to predict cell type from input data, which can then be saved and used for other data sets, even where cell type is not known.