Within-sibship genome-wide association analyses decrease bias in estimates of direct genetic effects

Area of Science:

• Human genetics
• Complex trait analysis
• Population genetics

Background:

• Genome-wide association studies (GWAS) typically use unrelated individuals, capturing direct genetic effects, demography, and indirect genetic effects from relatives.
• Family-based GWAS designs can better control for demographic and indirect genetic effects, but large-scale family datasets have been limited.

Purpose of the Study:

• To leverage large-scale sibling data to differentiate between population-level and within-family genetic effects for various human phenotypes.
• To assess the impact of family-based designs on estimates of heritability, genetic correlations, and causal inference using Mendelian randomization.

Main Methods:

• Combined data from 178,086 siblings across 19 cohorts to perform both between-family (population) and within-family (within-sibship) GWAS for 25 phenotypes.
• Compared GWAS estimates, SNP heritability, genetic correlations, and Mendelian randomization results between population and within-sibship analyses.
• Investigated evidence of polygenic adaptation using within-sibship estimates.

Main Results:

• Within-sibship GWAS estimates were smaller than population estimates for complex traits like height, educational attainment, and cognitive ability, indicating substantial influence of non-inherited factors.
• Differences were observed in downstream analyses, including attenuation of the genetic correlation between educational attainment and BMI towards zero in within-sibship analyses.
• Molecular phenotypes, such as low-density lipoprotein cholesterol, showed consistent results between population and within-sibship analyses, suggesting stronger direct genetic influence.

Conclusions:

• Family-based GWAS designs are crucial for accurately estimating direct genetic effects, particularly for phenotypes influenced by demographic factors and indirect genetic effects from relatives.
• Within-sibship analyses revealed smaller genetic effects for many complex human traits, highlighting the importance of accounting for shared environmental and familial influences.
• Within-sibship data provided evidence for polygenic adaptation, demonstrating the utility of family designs in evolutionary genetic studies.