The limits of normal approximation for adult height

Area of Science:

• Quantitative genetics
• Human genetics
• Biometrical studies

Background:

• Adult height was foundational for early biometrical and quantitative genetic studies.
• It led to the classical polygenic model, influencing complex trait analysis.
• This model assumes additive gene effects and normally distributed residuals.

Purpose of the Study:

• To evaluate the classical polygenic model's assumptions with large-scale adult height data.
• To explore necessary model adjustments for analyzing hundreds of thousands of individuals.
• To investigate alternative approximations for complex trait analysis.

Main Methods:

• Analysis of large-scale adult height data from hundreds of thousands of individuals.
• Comparison of the classical additive model with models incorporating non-additive interactions.
• Application of log-normal approximation as an alternative to normal distribution assumptions.

Main Results:

• The classical additive model with normal distribution assumptions proved insufficient for large datasets.
• Increased model complexity, including non-additive gene-environment-sex interactions, was required.
• Log-normal approximation provided a viable alternative, maintaining the additive effects model.

Conclusions:

• Standard polygenic models require refinement for massive genetic studies.
• Non-additive interactions are crucial for accurately modeling complex traits like height in large cohorts.
• Log-normal approximation offers a practical method for analyzing large-scale human height data.