New Method for Calculating Genetic Risk Scores
Author Information
Author(s): Huang Yu-Jyun, Kurniansyah Nuzulul, Goodman Matthew O, Spitzer Brian W, Wang Jiongming, Stilp Adrienne, Laurie Cecelia, de Vries Paul S, Chen Han, Min Yuan-I, Sims Mario, Peloso Gina M, Guo Xiuqing, Bis Joshua C, Brody Jennifer A, Raffield Laura M, Smith Jennifer A, Zhao Wei, Rotter Jerome I, Rich Stephen S, Redline Susan, Fornage Myriam, Kaplan Robert, Franceschini Nora, Levy Daniel, Morrison Alanna C, Boerwinkle Eric, Smith Nicholas L, Kooperberg Charles, Psaty Bruce M, Zöllner Sebastian, Sofer Tamar
Primary Institution: Harvard Medical School
Hypothesis
Can a new framework for polygenic risk scores (PRS) improve risk assessment across diverse populations?
Conclusion
The expected polygenic risk score (ePRS) framework provides a more equitable way to quantify genetic risk across diverse populations.
Supporting Evidence
- The ePRS framework can adjust for population stratification in association analysis.
- Simulation studies confirm that adjusting for ePRS yields nearly unbiased estimates of PRS-outcome associations.
- The ePRS framework was applied to analyze multiple cardiovascular-related traits.
- Results from the TOPMed dataset were consistent with those from the All of Us dataset.
- The ePRS framework allows for equitable risk classification across diverse populations.
- Adjustment for ePRS is more intuitive than using principal components for ancestry adjustment.
- The rPRS distribution is homogeneous across different populations, indicating effective calibration.
Takeaway
This study introduces a new way to calculate genetic risk that works better for people from different backgrounds, making it fairer for everyone.
Methodology
The study used simulation studies and applied the ePRS framework to analyze data from the TOPMed and All of Us datasets.
Potential Biases
Potential biases may arise from inaccuracies in estimating global and local ancestry.
Limitations
The accuracy of the ePRS framework depends on the precision of ancestry inference and allele frequency estimation.
Participant Demographics
Participants included up to 49,626 individuals from diverse ancestral backgrounds.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website