Comparing Methods to Detect Genetic Effects and Gene-Environment Interactions
Author Information
Author(s): Kazma Rémi, Dizier Marie-Hélène, Guilloud-Bataille Michel, Bonaïti-Pellié Catherine, Génin Emmanuelle
Primary Institution: Université Paris-Sud, UMR-S 535, Villejuif, France
Hypothesis
What is the power of different methods to detect the effect of Locus B and its interaction with tobacco use?
Conclusion
The study found that association studies are more powerful than linkage tests for detecting genetic effects and gene-environment interactions.
Supporting Evidence
- The mean interaction test had only 6% power to detect genetic effects.
- The log-linear modeling approach had a power of 78% for detecting genetic effects.
- The case-control design improved power to 95% when accounting for interaction.
- The case-only design reached 95% power for detecting G × E interaction.
- Type I error rates were higher than expected for all methods.
Takeaway
This study looked at different ways to find out how genes and smoking work together to affect health, and found that some methods are better than others.
Methodology
The study compared four methods using simulated data from Genetic Analysis Workshop 15, including a mean interaction test and three association testing methods.
Potential Biases
The case-only design's validity depends on the independence of genetic and environmental factors.
Limitations
The mean interaction test had poor power, and type I error rates were inflated for all methods.
Participant Demographics
Participants included affected sib pairs and case-parent triads, with a focus on tobacco use and rheumatoid arthritis risk.
Statistical Information
Confidence Interval
[1.03–1.71], [1.05–1.72]
Statistical Significance
p<0.05
Want to read the original?
Access the complete publication on the publisher's website