Bayesian Models for Pooling Microarray Studies
Author Information
Author(s): Conlon Erin M, Song Joon J, Liu Jun S
Primary Institution: Department of Mathematics and Statistics, University of Massachusetts, Amherst, Massachusetts, USA
Hypothesis
Can a Bayesian hierarchical model effectively pool cDNA microarray data from multiple independent studies to identify differentially expressed genes?
Conclusion
The Bayesian hierarchical model provides a systematic framework for combining microarray studies, leading to more accurate identification of differentially expressed genes.
Supporting Evidence
- The model produced significant increases in the number of discovered genes when pooling data from multiple studies.
- Bayesian estimates of false discovery rates were shown to track true FDRs effectively.
- The pooled model identified more differentially expressed genes than individual studies.
Takeaway
Scientists can combine results from different experiments to find more genes that are important for understanding diseases.
Methodology
A Bayesian hierarchical model was used to pool data from multiple cDNA microarray studies, allowing for the identification of differentially expressed genes.
Potential Biases
Potential biases may arise from the assumption of independence among studies and the requirement for data normalization.
Limitations
The model assumes that studies are independent and from the same platform, and it does not account for outliers.
Statistical Information
P-Value
p<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