AI-based characterization of Alzheimer's disease phenotypes from population-scale single-cell data.

AI-based framework identifies approximately 1.5 million phenotype-associated cells from over 1,000 brains with Alzheimer's disease phenotypes, revealing key cell subpopulations and pathways.

• Why it matters: Understanding the cellular and molecular basis of diverse AD phenotypes is crucial for developing targeted diagnostics and therapies, yet remains largely unclear.
• What they did: The study used a population-scale single-nucleus RNA sequencing dataset and developed PASCode, a graph neural model that scores single-cell phenotype associations across 584 donors, identifying relevant cell subpopulations and genes.
• The result: Findings highlight specific microglia, astrocyte, and neuronal subpopulations linked to AD phenotypes, and the framework, validated externally, offers a resource for precision medicine and community research tools.