Genetic association analysis using sibship data: a multilevel model approach.

Area of Science:

• Genetics
• Biostatistics
• Epidemiology

Background:

• Family-based association studies (FBAS) control for population stratification.
• FBAS can simultaneously test for genetic linkage and association.
• Analyzing sibship data presents unique statistical challenges.

Purpose of the Study:

• To propose and evaluate a novel retrospective multilevel model (rMLM) for analyzing sibship data.
• To compare the performance of rMLM against existing methods like S-TDT, SDT, CLR, and GEE.
• To assess the utility of rMLM in detecting gene-disease associations.

Main Methods:

• Developed a retrospective multilevel model (rMLM) treating genotypic information as the dependent variable.
• Generated simulated datasets using the SIMLA program.
• Compared rMLM with S-TDT, SDT, CLR, and GEE based on power, type I error, bias, and standard error.

Main Results:

• rMLM demonstrated validity for testing association in the presence of linkage using sibship data.
• rMLM showed enhanced advantages with concordant sibships.
• rMLM exhibited less underestimated odds ratios (OR) compared to GEE.
• rMLM is a robust method for gene-disease association detection in sibship data.

Conclusions:

• rMLM is a valuable tool for detecting gene-disease associations in sibship data.
• Caution is advised regarding potential increases in type I error rate when linkage is absent but association exists.
• The study supports the application of rMLM for genetic association studies.