Simultaneous selection of multiple important single nucleotide polymorphisms in familial genome wide association...

Area of Science:

• Genetics
• Statistical Genetics
• Bioinformatics

Background:

• Current single nucleotide polymorphism (SNP) association analysis is computationally intensive, often testing one SNP at a time.
• Joint modeling of multiple SNPs within genes or pathways offers greater power for detecting weakly associated variants.
• Familial dependency structures are often ignored in traditional model selection, limiting detection power.

Purpose of the Study:

• To develop a computationally efficient, resampling-based variable selection technique for identifying relevant SNPs in multi-marker mixed effect models.
• To enable simultaneous utilization of information from multiple SNPs for robust single SNP detection in family data.
• To improve the power of detecting genetic variants associated with complex traits.

Main Methods:

• A resampling-based approach utilizing the e-values framework for efficient model selection.
• Training a single model and employing a fast, scalable bootstrap procedure to overcome computational bottlenecks.
• Application of the method to analyze gene-level associations in the Minnesota Center for Twin and Family Research (MCTFR) dataset.

Main Results:

• The proposed method demonstrates superior effectiveness in detecting trait-associated SNPs compared to single-marker analysis in family data.
• It outperforms model selection methods that disregard familial dependency structures.
• Gene-level analysis identified several SNPs associated with alcohol consumption in the MCTFR dataset.

Conclusions:

• The developed resampling technique provides a computationally efficient and powerful tool for SNP detection in family-based genetic studies.
• This approach enhances the ability to identify genetic variants underlying complex traits by leveraging multi-marker information.
• The method has practical applications in large-scale genetic research, such as the analysis of alcohol consumption phenotypes.