Genome-wide association studies for discrete traits.

Area of Science:

• Genetics
• Statistical Genetics
• Bioinformatics

Background:

• Genome-wide association studies (GWAS) typically employ basic statistical tests like chi-squared or logistic regression for discrete traits.
• These methods may lack statistical power, particularly for complex genetic architectures.

Framework:

• Exploring advanced analytical approaches to increase the power of GWAS.
• Investigating methods that combine information from multiple genetic markers within a genomic region.
• Utilizing Bayesian variable selection for simultaneous genome-wide multivariate analyses.

Implementation:

• Methods explored include sliding windows, wavelets, Bayesian shrinkage, and penalized likelihood.
• Imputation techniques were assessed for handling missing genetic markers in individual studies and meta-analyses.
• The impact of single versus multiple imputation on association test robustness was evaluated.

Implications:

• Accurate control of population stratification is essential, as demonstrated by the disappearance of previously reported gene associations.
• Advanced methods offer potential for more powerful genetic discovery in complex diseases.
• Further research into subgroup analysis, gene-gene interactions, and biomarker utilization is warranted.