omicverse-single-cell-cellmatch-ontology

To map free-text cell-type labels to Cell Ontology terms, use CellOntologyMapper on your AnnData object. It applies sentence-transformer cosine similarity matching to translate author-specific annotations into standardized CL terms.

The best way to standardize single-cell annotations across cohorts is using species- or tissue-aware mapping. This approach leverages Cell Taxonomy context alongside CL resources to harmonize labels across mouse, human, or tissue-specific cohorts.

Yes, you can expand abbreviation-heavy cell labels before mapping. The Skill supports optional LLM abbreviation expansion to resolve shorthand labels and acronyms, ensuring accurate sentence-transformer cosine matching against canonical Cell Ontology terms.

Yes, this cell ontology mapping works with mouse and human tissue-specific datasets. It supports species and tissue enrichment by adding Cell Taxonomy context when standard CL resources are not enough for accurate cell-type standardization.

To handle ambiguous cell names during single-cell annotation harmonization, use configurable thresholding with local ontology embeddings. This allows you to inspect top matches for underspecified labels and filter low-confidence mappings effectively.