SNP selection for association studies: maximizing power across SNP choice and study size

Area of Science:

• Genetics
• Bioinformatics
• Statistical Genomics

Background:

• Selecting single nucleotide polymorphisms (SNPs) is crucial for genetic association studies.
• Current methods lack clear guidance on optimal SNP numbers and their relation to linkage disequilibrium (LD).
• SNP selection is often decoupled from sample size determination, despite resource limitations.

Purpose of the Study:

• To develop an optimization framework for SNP selection that considers both study size and LD patterns.
• To maximize the statistical power of association studies under budget constraints.
• To investigate the tradeoff between sample size and the number of SNPs to genotype.

Main Methods:

• Formulating SNP selection as an optimization problem to maximize study power within a budget.
• Employing a genetic algorithm and a hill climbing search to solve the optimization problem.
• Evaluating algorithm performance across different chromosomal regions with varying LD patterns.

Main Results:

• The proposed optimization approach effectively tunes SNP density to LD patterns.
• Both genetic and hill climbing algorithms efficiently identify optimal SNP sets.
• Selected SNP density significantly varies across chromosomal regions, reflecting underlying LD structures.

Conclusions:

• An integrated approach to SNP selection and sample size determination is feasible and beneficial.
• Optimization strategies can effectively adapt SNP selection to regional genomic characteristics.
• The developed algorithms provide efficient solutions for complex SNP selection problems in genetic association studies.