Choosing Null Distributions for Gene-Gene Interactions in GWAS
Author Information
Author(s): Yang Can, Wan Xiang, He Zengyou, Yang Qiang, Xue Hong, Yu Weichuan
Primary Institution: Hong Kong University of Science and Technology
Hypothesis
How does the choice of null distributions affect the detection of gene-gene interactions in genome-wide association studies?
Conclusion
Using permutation tests or independent data sets for hypothesis testing leads to more reliable results in detecting gene-gene interactions.
Supporting Evidence
- The study demonstrates that screening and modeling methods can change the null distribution used in hypothesis testing.
- Using permutation tests provides a more accurate estimation of null distributions.
- Testing on independent data sets can reserve the theoretical null distribution.
Takeaway
This study shows that picking the right way to compare genes is really important, or else we might think some genes are related to diseases when they aren't.
Methodology
The study used simulation experiments and real data analysis to evaluate the effects of different null distributions on hypothesis testing in GWAS.
Potential Biases
The study highlights risks of bias due to inappropriate null distribution choices leading to false positives or negatives.
Limitations
The study's findings may not generalize to all types of genetic studies, and the computational cost of permutation tests can be high.
Participant Demographics
The AMD study involved 96 cases and 50 controls.
Statistical Information
P-Value
4.226 × 10–7
Statistical Significance
p<0.05
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website