Single Marker Family-Based Association Analysis Not Conditional on Parental Information

Area of Science:

• Statistical Genetics
• Genomic Association Studies
• Bioinformatics

Background:

• Family-based association studies are crucial for identifying genetic variants influencing complex traits.
• Traditional methods often struggle with computational demands in genome-wide association studies (GWAS).
• Accounting for familial correlations is essential for robust genetic association analysis.

Purpose of the Study:

• To review and compare popular unconditional association analysis methods for family data.
• To discuss approaches for handling computational burdens in large-scale genome-wide association studies.
• To demonstrate practical application using R and S.A.G.E. software packages.

Main Methods:

• Marginal model using generalized estimating equations (GEE) to account for familial correlations.
• Mixed model incorporating a polygenic random component to model shared genomic effects.
• Review of genome-wide association analyses and computational strategies for large datasets.
• Demonstration of single-marker analysis using R package gee and S.A.G.E. ASSOC program.

Main Results:

• Unconditional methods provide greater statistical power compared to conditional approaches.
• GEE is effective when correlation structure is not of primary interest.
• Mixed models explicitly account for familial correlations via random effects.
• The ASSOC program offers flexibility for complex models and non-normal data in large pedigrees.

Conclusions:

• Unconditional association analysis methods offer a flexible framework for diverse pedigree data and trait distributions.
• Efficient computational strategies are necessary for applying these methods to genome-wide association studies.
• Software packages like R (gee) and S.A.G.E. (ASSOC) facilitate practical implementation and interpretation.