Dissecting the Predictive Accuracy of Polygenic Indexes for Behavioral Phenotypes Across Genetic Ancestries

Area of Science:

• Genetics
• Bioinformatics
• Population Genetics

Background:

• Polygenic indexes (PGIs) are widely used for genetic risk prediction.
• PGIs trained on European ancestry data exhibit reduced portability to non-European ancestries.
• The portability of PGIs for behavioral and social traits is less understood.

Purpose of the Study:

• To systematically analyze PGI portability across diverse ancestries for 52 health, behavioral, and social phenotypes.
• To investigate factors contributing to PGI accuracy loss in non-European populations.
• To compare PGIs derived from standard versus family-based Genome-Wide Association Studies (GWAS).

Main Methods:

• Utilized UK Biobank and Health and Retirement Study datasets.
• Calculated genome-wide PGIs.
• Assessed cross-ancestry heritability differences.
• Compared PGI performance using standard and family-based GWAS.

Main Results:

• Confirmed significant PGI predictive power reduction in non-European ancestries (African 24%, East Asian 37%, South Asian 51%).
• Biologically proximal traits showed greater portability than behavioral/social traits.
• Linkage disequilibrium and allele frequency differences were major drivers of PGI loss in African ancestries.

Conclusions:

• PGI portability is a significant issue, particularly for non-European ancestries and complex traits.
• Factors contributing to PGI portability loss vary across ancestries.
• Family-based GWAS PGIs offer modest improvements for specific traits, suggesting potential for mitigating portability gaps.