Bayesian t-test for Microarray Data
Author Information
Author(s): Richard J. Fox, Matthew W. Dimmic
Hypothesis
Can a two-sample Bayesian t-test improve the detection of differentially expressed genes compared to traditional methods?
Conclusion
The proposed Bayesian t-test shows equal or greater power than previous methods and maintains consistent Type I error rates.
Supporting Evidence
- The Bayesian t-test provides a more consistent approach to estimating variance.
- Simulations show that the new method has better power in detecting differentially expressed genes.
- The method is applicable beyond microarray data to any situation with prior variance information.
Takeaway
This study introduces a new way to compare gene expressions using a Bayesian approach, which is better at finding differences than older methods.
Methodology
The study proposes a Bayesian t-test that calculates the marginal distribution for the difference in mean expression without needing point estimates of variance.
Potential Biases
Potential biases may arise from the choice of prior distributions and the assumptions about variance.
Limitations
The method may not perform well with very small sample sizes or when prior information is not reliable.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website