celltype-annotation-scANVI

To annotate spatial transcriptomics cells, this Skill uses scANVI to transfer cell type labels from a labeled scRNA-seq reference dataset to your spatial data. It leverages scvi-tools and scanpy to train a model and generate per-cell type predictions for spatial data.

scANVI transfer learning for cell-type annotation is a semi-supervised approach that uses a labeled scRNA-seq dataset to predict cell types in an unlabeled spatial transcriptomics dataset. It reduces manual annotation effort by transferring known labels across datasets.

Yes, matching gene names between your scRNA-seq and spatial datasets is required for scANVI annotation. The process relies on common gene matching across datasets, combined with highly variable gene selection, to ensure accurate cross-dataset integration.

Yes, you can use scANVI to handle batch effects when integrating scRNA-seq and spatial transcriptomics data. The Skill supports batch handling during model training and inference to improve cross-dataset integration and cell-type prediction accuracy.

Limitations of using scANVI for spatial transcriptomics annotation include the strict requirement for a labeled reference scRNA-seq dataset and exact gene name matching. Without these prerequisites, the transfer learning model cannot produce per-cell type predictions.