EndoPRS: Incorporating Endophenotype Information to Improve Polygenic Risk Scores for Clinical Endpoints.

Area of Science:

• Genetics
• Bioinformatics
• Computational Biology

Background:

• Polygenic risk score (PRS) prediction for complex diseases can be enhanced by incorporating information from genetically correlated traits.
• Existing multi-trait PRS methods often fail to account for vertical pleiotropy, where one trait mediates another.

Purpose of the Study:

• To introduce endoPRS, a novel weighted lasso model designed to improve disease risk prediction by leveraging endophenotypes.
• To assess the performance of endoPRS across diverse complex genetic architectures and compare it with existing PRS methods.

Main Methods:

• Development of endoPRS, a weighted lasso model that integrates endophenotype data for disease risk prediction.
• Extensive simulation analyses to evaluate the robustness of endoPRS under various genetic models.
• Application of endoPRS to predict childhood-onset asthma risk using UK Biobank data and eosinophil count GWAS.

Main Results:

• endoPRS demonstrates robustness across a variety of complex genetic frameworks in simulation studies.
• Application to childhood-onset asthma prediction in UK Biobank showed significant improvement over existing PRS methods.
• endoPRS outperformed multi-trait PRS methods like MTAG and wMT-BLUP in predictive accuracy.

Conclusions:

• endoPRS offers a robust and effective approach for improving complex disease risk prediction by incorporating endophenotypes.
• The method's ability to handle vertical pleiotropy provides advantages over current multi-trait PRS techniques.
• endoPRS shows significant potential for real-life clinical applications in personalized medicine.