Comparative Analysis of Methods for Detecting Interacting Loci
Author Information
Author(s): Chen Li, Yu Guoqiang, Langefeld Carl D, Miller David J, Guy Richard T, Raghuram Jayaram, Yuan Xiguo, Herrington David M, Wang Yue
Primary Institution: Virginia Polytechnic Institute and State University
Hypothesis
A rigorous, comprehensive comparison of performance and limitations of available interaction detection methods is warranted.
Conclusion
The study provides new insights into the strengths and limitations of current methods for detecting interacting loci.
Supporting Evidence
- The best-performing method was MECPM.
- Most methods miss many interacting SNPs at an acceptable rate of false positives.
- Power varies for different models as a function of penetrance, minor allele frequency, and linkage disequilibrium.
Takeaway
This study compares different methods to find how genes interact with each other, helping scientists understand diseases better.
Methodology
The study compared eight methods on simulated data sets consistent with complex disease models, assessing detection power, type I error rate, and computational complexity.
Potential Biases
Different methods have varying biases based on their detection principles and statistical approaches.
Limitations
The study's methods may not generalize well to real-world data due to the complexity of interactions and the conservative nature of statistical assessments.
Participant Demographics
Simulated data sets based on 2000 subjects with SNPs from the NYCCCP.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website