This innovative research introduces a transcriptome foundation model that incorporates cell ontology knowledge to better understand gene expression patterns.

• Integrates hierarchical relationships between cell types that previous models ignored
• Demonstrates superior performance in cell type classification and perturbation prediction
• Achieves breakthrough capabilities in interpreting cancer drug responses
• Maintains high accuracy even with limited training data

This advancement matters for medicine by enabling more precise understanding of disease mechanisms and potential drug responses at the cellular level, potentially accelerating personalized treatment development.

Cell-ontology guided transcriptome foundation model