Modeling Gene Expression Data with a New Statistical Model
Author Information
Author(s): Helene H. Thygesen, Aeilko H. Zwinderman
Primary Institution: Academisch Medisch Centrum, University of Amsterdam
Hypothesis
Can a hierarchical Poisson model effectively separate sampling variance from biological variance in SAGE data?
Conclusion
The hierarchical Poisson model can effectively separate sampling variance from biological variance in gene expression data.
Supporting Evidence
- The model predicts a large number of zero counts, which cannot be observed.
- A log-normal prior provided a better fit to the data than the gamma prior.
- The model allows for the separation of sampling variance from biological variance.
Takeaway
This study created a new way to analyze gene expression data that helps scientists understand how much of the variation in gene expression is due to sampling errors versus actual biological differences.
Methodology
The study used a hierarchical Poisson model with a gamma prior and various algorithms to estimate parameters from SAGE data.
Potential Biases
Potential bias due to the model's inability to reliably estimate the shape parameter for low-expressed tags.
Limitations
The model struggles with estimating the number of zero counts and may not account for a small population of highly expressed tags.
Participant Demographics
The study analyzed data from 72 libraries from the Human Transcriptome Map, representing various human tissues.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website