Single-marker and multi-marker mixed models for polygenic score analysis in family-based data.

Area of Science:

• Genetics
• Statistical Genetics
• Bioinformatics

Background:

• Genome-wide association studies (GWAS) are powerful but limited in detecting variants with weaker effects.
• Aggregate association testing combines effects of multiple single-nucleotide polymorphisms (SNPs) to enhance power.
• This approach utilizes SNPs identified via single-marker or multi-marker modeling.

Purpose of the Study:

• To evaluate the performance of aggregate association testing using SNPs from different modeling strategies.
• To assess the impact of SNP set size and modeling approach on association replication in pedigree data.
• To identify challenges in assessing statistical significance for aggregate association tests in complex family structures.

Main Methods:

• Utilized simulated pedigree data from the Genetic Analysis Workshop 18.
• Performed quantitative trait association analysis for diastolic blood pressure and a control trait (Q1).
• Compared aggregate scores derived from single-marker and multi-marker SNP selection methods.

Main Results:

• The probability of replicating associations using aggregate scores was influenced by SNP set size.
• For smaller SNP sets (≤100), the modeling approach (single-marker vs. multi-marker) also affected replication probability.
• Assessing statistical significance proved difficult due to linkage disequilibrium in pedigree data and identical genotypes across replicates.

Conclusions:

• Aggregate association testing shows promise but requires careful consideration of SNP set composition and size.
• Challenges in statistical significance assessment necessitate the development of new methods for applying aggregate testing to pedigree data.
• Further research is needed to optimize aggregate association strategies for complex genetic architectures and family-based studies.