BRAINCELL-AID: An Agentic AI Created Brain Cell Type Resource for Community Annotation

BRAINCELL-AID is a multi-agent AI system that combines large language models with retrieval-augmented generation to accurately annotate brain cell types from single-cell RNA sequencing data. The tool achieved 77% accuracy on gene set annotations and successfully annotated 5,322 brain cell clusters from the mouse brain cell atlas, creating a community resource for cell type identification.

BRAINCELL-AID addresses a fundamental bottleneck in computational biology: the challenge of annotating single-cell transcriptomic data when gene function remains poorly characterized. Traditional methods like Gene Set Enrichment Analysis rely on pre-existing, often incomplete biological knowledge, while raw language models frequently hallucinate or misrepresent complex biological concepts. This system bridges that gap by combining the flexibility of LLMs with structured biological ontologies and literature grounding through retrieval-augmented generation, fundamentally improving annotation reliability.

The research demonstrates a broader trend in AI adoption within life sciences: moving beyond simple model application toward specialized agentic systems that incorporate domain-specific constraints and knowledge bases. The 77% accuracy rate on mouse gene sets represents meaningful progress for a notoriously difficult annotation task, particularly when considering that poorly characterized genes historically had near-zero annotation success rates.

The practical impact extends across neuroscience and cell biology research communities. By successfully annotating 5,322 brain cell clusters and identifying region-specific gene co-expression patterns, BRAINCELL-AID enables researchers to form new hypotheses about cell function without spending months on manual curation. The identification of Basal Ganglia-related cell types with neurologically meaningful descriptions validates the system's interpretability—a critical requirement for biological discovery tools.

Looking forward, this work signals how specialized AI systems optimized for scientific knowledge representation may become essential infrastructure for large-scale biological data analysis. As single-cell sequencing datasets proliferate, similar agentic approaches could accelerate annotation across multiple tissue types and organisms, potentially transforming how biomedical researchers approach high-dimensional biological data.

• →BRAINCELL-AID combines LLMs with retrieval-augmented generation to achieve 77% accuracy on gene set annotations, solving a major bottleneck in single-cell RNA sequencing analysis
• →The system successfully annotated 5,322 brain cell clusters from the comprehensive mouse brain cell atlas, identifying region-specific patterns and functional gene roles
• →Multi-agent AI workflows integrated with PubMed literature reduce hallucinations and improve interpretability compared to traditional bioinformatics methods
• →The tool creates a community resource for cell type annotation, accelerating biological discovery by reducing manual curation requirements
• →Specialized agentic systems optimized for domain knowledge represent an emerging pattern in AI adoption within life sciences research