spatial-annotate

Spatial transcriptomics cell type annotation assigns labels to tissue spots using methods like marker-based scoring, Tangram, scANVI, and CellAssign. It requires AnnData objects with spatial coordinates and optionally a single-cell RNA-seq reference dataset for label transfer.

Label transfer from single-cell RNA-seq to spatial data maps reference identities onto tissue spots. Tangram and scANVI use the single-cell reference to align molecular profiles, generating annotated spatial data matrices and visualizations of cellular distributions.

A single-cell reference dataset is not required for spatial transcriptomics annotation. While Tangram and scANVI leverage reference data for label transfer, the marker-based scoring method annotates spots directly using predefined marker genes.

Tangram and scANVI are supported annotation methods that can be applied to spatial data. Tangram performs mapping integration, while scANVI transfers labels probabilistically, offering flexibility based on available reference data and desired accuracy.

Annotating spatial transcriptomics cell types generates annotated data matrices, spatial maps, UMAP visualizations, and summary reports. These outputs visualize the distribution of identified cell types across the tissue architecture.

Spatial transcriptomics annotation requires spatial coordinates in the AnnData object to map assigned cell type labels back onto the physical tissue architecture. Without coordinates, the skill cannot generate spatial maps showing cellular distributions.