Evaluating Treatment Effects on Gene Groups
Author Information
Author(s): Robert Delongchamp, Taewon Lee, Cruz Velasco
Primary Institution: Division of Biometry and Risk Assessment, National Center for Toxicological Research, Jefferson, Arkansas, USA
Hypothesis
Can we reliably assess the statistical significance of treatments on groups of genes using adjusted methods for correlation?
Conclusion
Reliable corrections for the effect of correlations among genes on the significance level of a GO term can be constructed for a one-sided alternative hypothesis.
Supporting Evidence
- Computer simulations demonstrated that correlations among genes invalidate many statistical methods commonly used to assign significance to GO terms.
- Meta-analysis methods for combining p-values were modified to adjust for correlation.
- The bias of naïve significance calculations can be greatly decreased although not eliminated.
Takeaway
When scientists look at how treatments affect groups of genes, they need to be careful because genes can influence each other. This study shows how to better measure these effects.
Methodology
The study uses computer simulations and meta-analysis methods to adjust for correlations among genes when assessing treatment effects.
Potential Biases
The naïve significance calculations can overstate the significance due to correlations among genes.
Limitations
The methods may not be applicable to very small sample sizes, and the correlation structure must be estimated from limited data.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website