integration-scanorama

Batch integration across multiple spatial transcriptomics slices is performed by learning a shared embedding using Scanorama. This process removes technical variation while preserving biological signal across slices.

Scanorama batch correction learns a shared embedding across multiple slices, storing the corrected results in adata.obsm['X_scanorama']. It also generates downstream joint UMAP visualization and Leiden clustering results.

Preparing slices for Scanorama integration requires loading at least two slices into the session. The tool then identifies common genes, normalizes the data, and runs the integration to generate a joint corrected embedding.

Yes, you can use Scanorama stitching to combine two or more slices to compare batch effects and biological variation. It generates a joint embedding that enables direct comparison across the corrected slices.

You do not need to manually identify common genes before running Scanorama. The integration process automatically collects slices, identifies common genes, normalizes data, and applies the batch correction.