Detecting rare haplotype association with two correlated phenotypes of binary and continuous types

Area of Science:

• Genetics and Bioinformatics
• Statistical Genetics
• Computational Biology

Background:

• Joint analysis of correlated phenotypes in genetic studies offers advantages over single-trait analyses, including increased statistical power and improved understanding of genetic etiology.
• Analyzing multiple discordant phenotypes (e.g., binary and continuous) presents significant statistical challenges for joint modeling.
• Discovering associations between rare genetic variants (haplotypes) and multiple discordant phenotypes remains an unmet methodological need.

Purpose of the Study:

• To develop and evaluate a novel statistical method for the joint association analysis of rare haplotypes with two discordant phenotypes (binary and continuous).
• To address the limitations of existing methods in handling multiple, mixed-type phenotypes for genetic association studies.
• To identify genetic variants contributing to complex traits like lung cancer and nicotine dependence.

Main Methods:

• Proposed the bivariate logistic Bayesian LASSO with a continuous outcome (bivariate LBL-BC) method for joint haplotype association analysis.
• Utilized a latent variable approach to model the correlation between binary and continuous phenotypes.
• Conducted extensive simulations to compare the performance of bivariate LBL-BC against univariate LBL and bivariate LBL-2B methods.

Main Results:

• The bivariate LBL-BC method demonstrated superior performance in most simulated scenarios compared to existing univariate and bivariate approaches.
• Performance was comparable only when phenotypes were weakly correlated and the haplotype affected only the binary trait, or when phenotypes were strongly positively correlated and the haplotype affected both positively.
• Application to lung cancer and nicotine dependence data successfully identified several associated haplotypes, including a rare one.

Conclusions:

• The bivariate LBL-BC method provides a powerful and effective tool for joint genetic association analysis of rare haplotypes with multiple discordant phenotypes.
• This approach enhances the ability to uncover complex genetic architectures underlying diseases.
• The method has practical utility, as demonstrated by its application to real-world data identifying potential genetic contributors to lung cancer and nicotine dependence.