Estimating the Proportion of True Null Hypotheses in Multiple Testing
Author Information
Author(s): Shu-Dong Zhang
Primary Institution: Centre for Cancer Research and Cell Biology (CCRCB), Queen's University Belfast
Hypothesis
What is the most accurate method for estimating the proportion of true null hypotheses in multiple testing?
Conclusion
The ZG04 method is identified as the most accurate for estimating the proportion of true null hypotheses across a wide range of conditions.
Supporting Evidence
- The ZG04 method consistently outperformed other methods in accuracy across a wide parameter space.
- Statistical tests showed significant differences in accuracy between the ZG04 method and others.
- The study covered an unprecedented range of conditions in its simulations.
Takeaway
This study looked at different ways to figure out how many of the tests we run are actually showing real results, and found that one method works best most of the time.
Methodology
The study compared five methods for estimating the proportion of true null hypotheses using extensive in silico experiments.
Potential Biases
The methods may not perform equally well under all conditions, particularly when the assumptions about independence do not hold.
Limitations
The study focused on a limited number of methods and used a single distributional assumption for gene expression.
Statistical Information
P-Value
p<0.005
Statistical Significance
p<0.005
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website