celltype-annotation-fast

Annotate cell types in spatial transcriptomics data by combining unsupervised Leiden clustering to identify cell populations, extracting cluster-specific marker genes, and using an LLM to assign precise cell type names based on tissue context.

Yes, LLM-based labeling rapidly annotates cell types on single-slice spatial transcriptomics datasets by mapping unsupervised clusters to specific cell populations using derived marker genes and provided tissue type context.

The best way to identify cellular composition is using a workflow that combines unsupervised clustering, marker gene analysis, and LLM-based labeling to provide fast and interpretable identification of cellular architecture within tissue sections.

Yes, you must provide a tissue type context, a spatial dataset, and an annotation function to run the preprocessing workflow that performs clustering, marker extraction, and mapping clusters to cell types.

Unsupervised Leiden clustering groups cells into populations, marker gene analysis derives specific markers for each cluster, and the LLM uses these markers alongside tissue context to assign precise cell type names for prediction.

This LLM-based annotation approach is applicable to single-slice datasets with gene expression and tissue context for exploratory discovery, meaning it is designed for rapid annotation rather than handling multi-slice or multi-sample integrated analyses.