Impact of Gene Correlation on Gene Set Testing
Author Information
Author(s): Gatti Daniel M, Barry William T, Nobel Andrew B, Rusyn Ivan, Wright Fred A
Primary Institution: University of North Carolina at Chapel Hill
Hypothesis
How does the correlation between genes affect the results of gene set testing methods?
Conclusion
The study reveals that traditional gene set testing methods often yield high false positive rates due to the correlation between genes, suggesting a need for resampling-based methods.
Supporting Evidence
- Correlation between genes inflates false positive rates in gene set testing.
- Resampling methods provide a more accurate assessment of statistical significance.
- High false positive rates were observed in traditional methods, often exceeding 50%.
Takeaway
When scientists look at groups of genes, they often think they are finding important patterns, but sometimes they are just seeing things that aren't really there because the genes are related to each other.
Methodology
The study conducted a meta-analysis of over 200 datasets from the Gene Expression Omnibus, comparing traditional gene set testing methods with resampling approaches.
Potential Biases
The reliance on independence assumption methods may lead to misinterpretation of results due to high false positive rates.
Limitations
The study primarily focuses on datasets with 20 or more samples, which may not represent smaller studies.
Participant Demographics
The datasets included a variety of experiments, including in vitro cell lines, sex comparisons, multiple cancer sub-types, tissues, and diseases.
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