Methods for Detecting Gene Interactions in Family Studies
Author Information
Author(s): Brock Guy N, Maher Brion S, Goldstein Toby H, Cooper Margaret E, Marazita Mary L
Primary Institution: University of Pittsburgh
Hypothesis
Can we effectively detect gene × gene interactions in large extended pedigrees?
Conclusion
The study found that detecting gene interactions is challenging, but the covariate-based approach showed more promise than the correlation-based method.
Supporting Evidence
- The correlation-based method showed a high type I error rate, indicating liberal significance levels.
- The covariate-based approach using LODPAL outperformed the correlation-based method in detecting interactions.
- Significant interactions were more easily detected when analyzing simpler models.
Takeaway
This study is about figuring out how genes work together in families to cause diseases, and it shows that it's really hard to find these interactions, but some methods work better than others.
Methodology
The study used simulated data to assess two methods for detecting gene interactions: a correlation-based approach and a covariate-based approach using LODPAL.
Potential Biases
The use of LODPAL may over-inflate linkage signals due to treating all affected relative pairs as independent.
Limitations
The methods struggled with detecting interactions in heterogeneous disease phenotypes and had high type I error rates.
Participant Demographics
Data were collected from four populations, including nuclear families and three-generation pedigrees.
Statistical Information
P-Value
0.05
Statistical Significance
p<0.05
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website