Immune atlas and annotation

Use this workflow when you need to identify immune populations, assign reference-atlas labels, and score immune states before downstream analysis. It is usually the first use case for single-cell or CITE-seq immune studies because communication, repertoire, disease, and reporting workflows all depend on credible cell labels.

Research question

Which immune populations and states are present, how confident are the reference-atlas labels, and which cells need review before downstream repertoire, communication, or disease workflows?

Use case

Use this when a dataset needs broad-to-fine immune labels, reference-label confidence, marker evidence, and state program scores. The workflow helps separate confident assignments from ambiguous or out-of-reference cells while preserving reference provenance for methods and reporting.

Suggested path

1. Register a scrna or cite_seq dataset.
2. Run immune_composition.
3. Optionally upload a reference atlas table with labels, broad labels, marker genes, organism, reference name, and version.
4. Run immune_annotation.
5. Review reference labels, confidence categories, marker overlap warnings, competing labels, and state programs.
6. Generate AI interpretation if completed runs are available.

Outputs to cite

• summary.total_cells
• summary.unique_immune_labels
• summary.unique_reference_labels
• summary.mean_reference_probability
• label counts
• reference label counts and confidence categories
• reference atlas compatibility warnings
• state program mean scores
• gene set and reference atlas source/version