Analysis of baseline, average, and longitudinally measured blood pressure data using linear mixed models

Area of Science:

• Genetics
• Biostatistics
• Genomics

Background:

• Genome-wide association studies (GWAS) are crucial for identifying genetic variants linked to complex traits.
• Analyzing longitudinal data in GWAS presents challenges due to relatedness and repeated measures.
• Linear mixed models (LMMs) offer a framework to account for these complexities.

Purpose of the Study:

• To compare the performance of three LMM-based methods for genetic variant identification in GWAS.
• To evaluate methods using baseline, average, and longitudinal data analysis approaches.
• To assess accuracy in identifying a known blood pressure-associated variant using simulated and real phenotype data.

Main Methods:

• Application of three LMM approaches: baseline measures, mean outcome measures, and longitudinal measurements.
• Inclusion of covariates as fixed effects and individual relatedness as a random effect variance-covariance structure.
• Utilizing the GRAMMAR approach for data decorrelation within LMMs.

Main Results:

• The LMM with longitudinal measurements demonstrated the highest accuracy in identifying a known single-nucleotide polymorphism associated with diastolic blood pressure.
• LMMs utilizing baseline measures performed best for systolic blood pressure identification.
• Both simulated and real phenotype data analyses were conducted to validate findings.

Conclusions:

• Longitudinal data analysis using LMMs is highly effective for identifying genetic variants influencing diastolic blood pressure in GWAS.
• Baseline data analysis with LMMs is a strong approach for systolic blood pressure-related variant discovery.
• Method selection should consider the specific trait and data structure for optimal GWAS performance.