An efficient method to handle the 'large p, small n' problem for genomewide association studies using Haseman-Elston...

Area of Science:

• Genetics
• Statistical Genomics
• Bioinformatics

Background:

• The 'large p, small n' problem poses a significant challenge in genomewide association studies (GWAS), where the number of genetic markers (p) far exceeds the number of samples (n).
• Existing methods for addressing this issue in GWAS have limitations, particularly in integrating regression techniques with scanning approaches.

Purpose of the Study:

• To extend the Haseman-Elston (H-E) regression method for application in GWAS.
• To develop a novel approach that combines H-E regression with sliding-window scan techniques.
• To improve statistical power in detecting quantitative trait loci (QTLs) within the context of GWAS.

Main Methods:

• Extended Haseman-Elston (H-E) regression for GWAS.
• Utilized phenotype measurements from sib pairs.
• Applied hidden Markov models to infer identity by state (IBS).

Main Results:

• The proposed H-E regression sliding-window approach demonstrated higher statistical power compared to single-marker association studies.
• The method successfully captured approximately 48.01% of quantitative trait loci (QTLs).
• Statistical power decreased with an increasing number of QTLs and was sensitive to heritability.

Conclusions:

• The integrated H-E regression and sliding-window scan method offers a powerful new tool for GWAS.
• This approach effectively addresses the 'large p, small n' problem by enhancing QTL detection.
• Further research should consider the impact of multiple QTLs and varying heritability on the method's performance.