A generalized least-squares framework for rare-variant analysis in family data

Area of Science:

• Genetics
• Statistical Genetics
• Bioinformatics

Background:

• Genome-wide association studies (GWAS) often fail to explain the full heritability of complex traits.
• Rare genetic variants are increasingly recognized as important contributors to missing heritability.
• Family-based study designs offer advantages for detecting rare causal variants due to enriched familial co-segregation.

Purpose of the Study:

• To evaluate the performance of the generalized least squares, sequence kernel association test (GLS-SKAT) method for rare variant association analysis in family samples.
• To compare the power and type I error rate of GLS-SKAT against existing methods using simulated family data.

Main Methods:

• Development of the GLS-SKAT approach, which utilizes a kinship matrix to account for familial relatedness and decorrelate family structures.
• Application of the SKAT-O (Omnibus) test for gene or region-based inference on the decorrelated data.
• Analysis of simulated systolic blood pressure data from the Genetic Analysis Workshop 18 family sample.

Main Results:

• The GLS-SKAT method demonstrated superior statistical power in detecting rare variant associations compared to the family-based association test (FBAT-v1).
• GLS-SKAT maintained good control over the type I error rate, indicating reliable results.
• The method effectively leveraged family structure to enhance the detection of rare causal variants.

Conclusions:

• GLS-SKAT is a powerful and reliable method for rare variant association studies in family-based samples.
• This approach can improve the identification of genetic variants contributing to complex traits, addressing limitations of current GWAS.
• The GLS-SKAT method provides a valuable tool for genetic analysis in families, enhancing the discovery of disease-related genetic factors.