Association mapping of blood pressure levels in a longitudinal framework using binomial regression

Area of Science:

• Genetics
• Biostatistics
• Cardiovascular Research

Background:

• Quantitative traits underpin complex genetic traits, but single phenotypes may not fully represent clinical endpoints.
• Existing association analysis methods (e.g., variance components, principal components) have limitations in robustness and biological interpretation.
• Multivariate phenotype vectors correlated with endpoint traits offer a potentially more optimal approach for association studies.

Purpose of the Study:

• To propose and evaluate a novel regression approach for detecting associations between single-nucleotide polymorphisms (SNPs) and multivariate phenotype vectors.
• To address limitations of existing methods in robustness and biological interpretability.
• To analyze genetic associations with systolic and diastolic blood pressure using a bivariate approach.

Main Methods:

• Development of a novel regression approach utilizing a conditional binomial model.
• Application of the proposed method to analyze Genetic Analysis Workshop 18 data.
• Comparison of bivariate analysis results with univariate analyses for blood pressure phenotypes.

Main Results:

• The proposed conditional binomial regression model effectively detects associations between SNPs and multivariate phenotype vectors.
• Bivariate analysis of systolic and diastolic blood pressure yielded more significant results (lower p-values) compared to univariate analyses.
• The novel method demonstrates potential for improved genetic association findings.

Conclusions:

• A novel conditional binomial regression approach offers a robust and interpretable method for multivariate phenotype association analysis.
• Analyzing multiple correlated phenotypes simultaneously, such as blood pressure, can enhance the power to detect genetic associations.
• This approach provides a promising alternative for studying complex genetic traits and their clinical manifestations.