CASSIA 0.3.2

CASSIA (Cell type Annotation using Specialized System with Integrated AI) is a Python package for automated cell type annotation in single-cell RNA sequencing data using large language models.

CASSIA

CASSIA is a Python and R package designed for automated, accurate, and interpretable single-cell RNA-seq cell type annotation using a modular multi-agent LLM framework. CASSIA provides comprehensive annotation workflows that incorporate reasoning, validation, quality scoring, and reporting—alongside optional agents for refinement, uncertainty quantification, and retrieval-augmented generation (RAG).

Highlights

• 🔬 Reference-free and interpretable LLM-based cell type annotation
• 🧠 Multi-agent architecture with dedicated agents for annotation, validation, formatting, quality scoring, and reporting
• 📈 Quality scores (0–100) and optional consensus scoring to quantify annotation reliability
• 📊 Detailed HTML reports with reasoning and marker validation
• 💬 Supports OpenAI, Anthropic, OpenRouter APIs and open-source models (e.g., LLaMA 3.2 90B)
• 🧬 Compatible with markers from Seurat (FindAllMarkers) and Scanpy (tl.rank_genes_groups)
• 🚀 Optional agents: Annotation Boost, Subclustering, RAG (retrieval-augmented generation), Uncertainty Quantification
• 🌎 Cross-species annotation capabilities, validated across human, mouse, and non-model organisms
• 🧪 Web UI also available: https://www.cassiacell.com

Installation

Install the core CASSIA framework:

pip install CASSIA

To enable optional RAG functionality:

pip install CASSIA_rag

Note: For R users, see the R package on GitHub.

Quick Start

# Run the CASSIA pipeline in fast mode
CASSIA.runCASSIA_pipeline(
    output_file_name = "FastAnalysisResults",
    tissue = "large intestine",
    species = "human",
    marker_path = unprocessed_markers,
    max_workers = 6,  # Matches the number of clusters in dataset
    annotation_model = "openai/gpt-4o-2024-11-20", #openai/gpt-4o-2024-11-20
    annotation_provider = "openrouter",
    score_model = "anthropic/claude-3.5-sonnet",
    score_provider = "openrouter",
    score_threshold = 75,
    annotationboost_model="anthropic/claude-3.5-sonnet",
    annotationboost_provider="openrouter"
)

For detailed workflows and agent customization, see the Example.

Contributing

We welcome contributions! Please submit pull requests or open issues via GitHub.

License

MIT License © 2024 Elliot Xie and contributors.

Support

Open an issue on GitHub or visit CASSIAcell.com for help.