A General Approach to Adjusting Genetic Studies for Assortative Mating

Area of Science:

• Genetics
• Quantitative genetics
• Behavioral genetics

Background:

• Assortative mating (AM), the non-random association of genotypes or phenotypes, increasingly impacts genetic study estimates.
• Existing AM theory requires complex, unknown historical sorting patterns for accurate adjustments.
• The influence of AM on genetic correlations and Mendelian randomization (MR) requires robust analytical methods.

Approach:

• Developed a general-purpose approach utilizing polygenic indexes (PGIs) to estimate AM's contribution to genetic variance and correlation.
• Extended existing AM theory to accommodate more general sorting models.
• Applied the PGI approach to UK Biobank data to quantify AM's effect on genetic correlations.

Key Points:

• The PGI approach can estimate the fraction of genetic variance and correlation attributable to AM.
• AM introduces selection bias in Mendelian randomization (MR) studies that persists even after adjustment.
• The PGI adjustment in MR is highly sensitive to PGI estimation errors.

Conclusions:

• Assortative mating (AM) inflates genetic correlation estimates between health and education traits by an average of 14% in UK Biobank data.
• The proposed PGI method offers a simpler, theory-based adjustment for AM effects in genetic studies.
• Researchers should exercise caution when interpreting genetic correlations or MR estimates for traits influenced by AM.