Skip to Main Content
MyYSM
YSM Home

INFORMATION FOR

This study introduces scPRS, a graph neural network framework that improves genetic risk prediction and links genetic variants to cell-specific disease mechanisms for conditions like Alzheimer's and diabetes.

Single-Cell PRS Framework Enhances Disease Risk Prediction

Publication Title: Single-cell polygenic risk scores dissect cellular and molecular heterogeneity of complex human diseases

Summary

Question
This study introduces single-cell polygenic risk scores (scPRS), a novel framework that integrates genetic risk prediction with single-cell chromatin accessibility data to dissect cellular and molecular heterogeneity in complex diseases. The authors aimed to improve the predictive accuracy of polygenic risk scores (PRS) while uncovering disease-critical cell types and gene regulatory mechanisms.
Why it Matters
Complex diseases like type 2 diabetes, Alzheimer’s disease, and hypertrophic cardiomyopathy involve multiple cell types and genetic variants, many of which are poorly understood. scPRS offers a method to not only predict genetic risk but also identify specific cells and gene regulatory programs linked to disease. This approach has the potential to advance precision medicine by improving diagnosis, risk stratification, and treatment targeting.
Methods
The researchers developed scPRS using graph neural networks (a type of machine learning model) to analyze genetic risk at the single-cell level. They incorporated chromatin accessibility data from single-cell assays to identify genetic variants and their impact on specific cell types. The method was applied to datasets for diseases such as type 2 diabetes, Alzheimer’s disease, hypertrophic cardiomyopathy, and severe COVID-19, with results validated using experimental and computational benchmarks.
Key Findings
scPRS outperformed traditional PRS methods in predicting genetic risk across multiple diseases. It successfully identified disease-critical cell types, such as pancreatic beta cells for type 2 diabetes and microglia for Alzheimer’s disease. The framework also linked specific genetic variants to disrupted gene regulation in these cells. For instance, scPRS revealed that a genetic variant in Alzheimer’s disease impacts the expression of genes involved in microglial phagocytosis, a key immune function.
Implications
The findings demonstrate that scPRS can bridge genetic risk with cell-specific biology, enabling more precise insights into disease mechanisms. This approach could inform targeted therapies by identifying the most relevant cell types and molecular pathways for intervention. Additionally, scPRS could improve genetic risk prediction, especially for diseases with significant cellular heterogeneity.
Next Steps
The authors propose expanding scPRS to integrate additional single-cell modalities, such as RNA sequencing and DNA methylation data, to capture broader layers of genetic regulation. Further studies are needed to validate scPRS findings in larger, more diverse populations and to explore its application in other complex diseases.
Funding Information
This research was supported by the National Institutes of Health (NIH) under awards CEGS 5P50HG007735, R01AG079291, RF1AG079557, 2R01NS097850, and 1R01NS131409. Additional funding was provided by the Tau Consortium and the John Douglas French Alzheimer’s Foundation. Support for this work also came from the Million Veteran Program, Office of Research and Development, Veterans Health Administration (MVP001).

Full Citation

Zhang S, Shu H, Zhou J, Rubin-Sigler J, Yang X, Liu Y, Cooper-Knock J, Monte E, Zhu C, Tu S, Li H, Tong M, Ecker J, Ichida J, Shen Y, Zeng J, Tsao P, Snyder M. Single-cell polygenic risk scores dissect cellular and molecular heterogeneity of complex human diseases. Nature Biotechnology 2025, 1-17. PMID: 40715455, DOI: 10.1038/s41587-025-02725-6.

Authors

  • Sai Zhang

    First Author
    Yale School of Medicine

    Assistant Professor of Biomedical Informatics and Data Science

    View Full Profile

  • Philip S. Tsao

    Last Author
    School Building Streamline Icon: https://streamlinehq.comOther Institution

  • Michael P. Snyder

    Last Author
    School Building Streamline Icon: https://streamlinehq.comOther Institution

Other Authors

  • Hantao Shu;
  • Jingtian Zhou;
  • Jasper Rubin-Sigler;
  • Xiaoyu Yang;
  • Yuxi Liu;
  • Johnathan Cooper-Knock;
  • Emma Monte;
  • Chenchen Zhu;
  • Sharon Tu;
  • Han Li;
  • Mingming Tong;
  • Joseph R. Ecker;
  • Justin K. Ichida;
  • Yin Shen;
  • Jianyang Zeng

Research Themes

Concepts