scvi-scanvi

To transfer cell-type annotations in scRNA-seq, you can use semi-supervised label propagation by providing seed labels and a trained scVI model to guide the annotation of new cells while preserving biological consistency.

Semi-supervised cell-type transfer in single-cell RNA-seq is a process that uses partial seed labels to propagate annotations across new datasets, leveraging a trained scVI model to maintain biological conservation during the transfer.

Yes, you need a trained scVI model to run scANVI for cell-type annotation. The scANVI workflow initializes from the trained scVI model to ensure biological consistency before propagating seed labels to unlabeled cells.

After running semi-supervised cell-type labeling, you obtain predicted cell-type labels and latent representations that are directly suitable for downstream clustering and visualization of your single-cell data.

Yes, you can use scANVI with partial seed labels for your scRNA-seq data. The model applies to scenarios with either partial or full label seeds to annotate new datasets effectively.

Using a trained scVI model for semi-supervised cell-type annotation preserves biological consistency across experiments, allowing seed labels to propagate more accurately to new datasets than fully supervised methods without this generative foundation.