API

Preprocessing

pp.bin_data(adata[, n_bins, layer])	Discretize expression values into n_bins per-cell quantile bins.
pp.slice_and_bin(adata, species[, binning, ...])	Subset to species-specific manual genes and bin the chosen layer.
pp.get_genes(species)	Return the manually curated gene list for a species.
pp.align_to_genes(adata, genes, *[, ...])	Subset/reindex an AnnData to a fixed gene panel and order.

Tools

tl.extend_atlas(query, atlas, species, *[, ...])	Map a query dataset onto the reference atlas via scArches surgery.
tl.embed_and_predict(adata, species, *[, ...])	Embed a query dataset with a packaged SCANVI model and predict cell types.
tl.predict_labels(adata, species, *[, ...])	Predict hierarchical Level-4 cell-type labels for a query dataset.
tl.HierarchicalClassifier(root, malignant, ...)	Wraps the three MLPs of the hierarchical classifier for one species.
tl.MLP(n_genes, n_classes[, hidden, dropout])	A simple feed-forward classifier with BatchNorm + dropout hidden blocks.
tl.derive_malignant_mask(level4)	Derive a boolean malignant mask from Level-4 labels.

Plotting

pl.plot_label_distribution(adata[, key, ax])	Plot a horizontal bar chart of the value counts of an obs column.
pl.basic_plot(adata)	Placeholder plotting function kept for template/back-compat purposes.

Models

models.list_models()	List the directory names of packaged ScanVI models.
models.available_models()	Return a mapping {alias -> directory name} for packaged ScanVI models.
models.load_scanvi_model(alias, *, adata[, ...])	Load a packaged ScanVI model by alias using scvi-tools' SCANVI.load.
models.load_classifier_checkpoints(species)	Load the three hierarchical-classifier checkpoints for a species.