Testing linkage and gene x environment interaction: comparison of different affected sib-pair methods

Area of Science:

• Genetic Epidemiology
• Statistical Genetics
• Bioinformatics

Background:

• Gene-environment (G x E) interactions are crucial for understanding complex diseases.
• Identifying G x E interactions requires robust statistical methods to analyze genetic linkage and interaction effects.

Purpose of the Study:

• To compare the power of different statistical tests in detecting linkage and G x E interaction using affected sib-pair data.
• To evaluate methods under various G x E interaction models.

Main Methods:

• Maximum Likelihood Lod-Score (MLS) based on allele sharing.
• Sum of MLS (sMLS) considering exposure status.
• Predivided Sample Test (PST) comparing allele sharing across exposure groups.
• Triangle Test Statistic (TTS) using discordant sib-pairs.
• Mean Interaction Test (MIT) regressing allele sharing on exposure.

Main Results:

• PST and TTS can detect both linkage and G x E interaction.
• PST, sMLS, and MIT may offer greater power for linkage detection than MLS when exposure modifies gene effects.
• TTS can be more powerful than other G x E-aware tests when exposure reverses gene effect direction.
• MLS remains most powerful for linkage detection when G x E effects are not strongly modifying.

Conclusions:

• The PST and TTS are valuable for detecting both genetic linkage and G x E interactions.
• The choice of test depends on the specific G x E model and whether interaction detection or linkage detection is prioritized.
• Understanding G x E interaction models is key to selecting appropriate statistical approaches.