A grid-search algorithm for optimal allocation of sample size in two-stage association studies.

Area of Science:

• Genetics
• Statistical Genetics
• Bioinformatics

Background:

• Genome-wide association studies (GWAS) involve extensive multiple testing due to dense single nucleotide polymorphism (SNP) mapping.
• Traditional family-wise error rate (FWER) controlling methods can be overly conservative and lack power in GWAS.
• Two-stage multiple testing strategies offer improved power and efficiency for detecting disease-associated SNPs.

Purpose of the Study:

• To develop an optimal sample size allocation strategy for two-stage GWAS.
• To address constraints of fixed overall cost and limited total sample size.
• To maximize statistical power while controlling false-positive rates in SNP association analysis.

Main Methods:

• Proposal of a grid-search algorithm for optimal sample size allocation in two-stage procedures.
• Consideration of two constraint types: fixed overall cost and limited total sample size.
• Simulation studies to evaluate the performance of the proposed allocation method.

Main Results:

• Optimal allocation, particularly allocating at least 80% of the total cost to stage one, generally maximizes power.
• Adjusting the proportion of cost in earlier stages can maintain good power when per-genotyping costs differ between stages.
• For limited sample sizes, evaluating all markers on 55% of subjects in stage one yields maximum power with ~43% cost reduction.

Conclusions:

• The proposed grid-search algorithm provides an effective method for optimizing sample size allocation in two-stage GWAS.
• Optimal sample size allocation is crucial for achieving a balance between statistical power, cost-efficiency, and error control.
• The findings offer practical guidelines for designing cost-effective and powerful GWAS.