omicverse-single-cell-annotation

What problem does it solve?

Turn clustered single-cell AnnData into reliable cell-type labels without manual label curation by providing a reusable, backend-selectable annotation workflow that encapsulates CellTypist, LLM-based mapping, and SCSA marker-database scoring.

Core Features & Use Cases

• Multiple backends: Choose CellTypist for pretrained model predictions, gpt4celltype for LLM-driven marker-to-label mapping, or SCSA for database-scoring annotation.
• Validated output contract: Writes branch-specific label columns and prediction matrices into AnnData and includes validation checks for cluster keys and annotation-ready inputs.
• Integration use case: Map leiden clusters from a preprocessed AnnData to cell-type labels for downstream visualization and differential analysis.

Quick Start

Annotate a clustered AnnData object by instantiating OmicVerse's Annotation with your AnnData and calling Annotation.annotate with method set to one of celltypist, gpt4celltype, or scsa.