Disease model distortion in association studies.

Area of Science:

• Genetics
• Statistical Genetics
• Bioinformatics

Background:

• Genome-wide association studies (GWAS) typically assume a single nucleotide polymorphism (SNP) follows a multiplicative disease model.
• Deviations from this model, such as dominance or interaction effects, are challenging to detect in GWAS.
• Imperfect linkage disequilibrium (LD) between causal variants and marker SNPs can distort observed genetic effects.

Purpose of the Study:

• To analytically and empirically quantify the impact of linkage disequilibrium on detecting departures from the multiplicative model in GWAS.
• To assess the power of GWAS to detect non-multiplicative genetic models under realistic LD patterns.

Main Methods:

• Analytical derivation of power decay as a function of linkage disequilibrium (r^4 for non-multiplicative models).
• Simulation studies using empirical LD patterns to evaluate disease model distortion in GWAS.
• Assessment of power to detect dominant and recessive models.

Main Results:

• Power to detect deviations from the multiplicative model decays rapidly as a function of r^4, where r^2 is the marker-causal locus correlation.
• Despite rapid decay, simulation studies show reasonable power to detect substantial deviations like dominant and recessive models among associated loci.
• Linkage disequilibrium significantly distorts genetic disease models.

Conclusions:

• Explicitly testing for deviations from the multiplicative model (e.g., dominant, recessive) in GWAS is warranted.
• Understanding LD effects is crucial for accurate genetic model inference in GWAS.
• Current GWAS methodologies may underestimate non-multiplicative genetic effects.