OmicVerse Single-Cell CytoTRACE2

Goal

Run the reusable CytoTRACE2 execution spine on single-cell AnnData: prepare the expression matrix, ensure pretrained weights are available, run ov.single.cytotrace2(...), capture the score and potency columns written into adata.obs, and optionally overlay those outputs on an existing embedding. Keep the skill centered on the potency-prediction job rather than on the tutorial dataset.

Quick Workflow

1. Inspect the input AnnData, especially whether adata.X contains the expression values you want CytoTRACE2 to score and whether the gene symbols match the declared species.
2. If the object still needs notebook-style normalization and HVG annotation, run ov.pp.preprocess(...) first and choose the mode branch explicitly.
3. Ensure the pretrained CytoTRACE2 weight directory is present before inference; download or stage it if needed.
4. Run ov.single.cytotrace2(...) with explicit species, batch settings, and parallelization settings.
5. Treat the returned table plus the five adata.obs columns as the shared handoff point for optional visualization.
6. Only if the user asks for plots, overlay CytoTRACE2_Score, CytoTRACE2_Potency, or CytoTRACE2_Relative on an existing embedding.
7. Validate the expected output columns and result table before treating the run as complete.

Interface Summary