Unified Gene Expression Measures from Multiple Platforms
Author Information
Author(s): Wang Xin Victoria, Verhaak Roel G. W., Purdom Elizabeth, Spellman Paul T., Speed Terence P.
Primary Institution: Dana-Farber Cancer Institute
Hypothesis
Can factor analysis be used to create a unified gene expression measure from multiple microarray platforms?
Conclusion
Factor analysis provides a more reliable estimate of gene expression by combining measurements from different platforms.
Supporting Evidence
- The unified expression measure is a weighted average of the observed expression measures from the three platforms.
- Factor analysis down-weights problematic platforms to improve accuracy.
- Gene-level FA and gene-level averages performed comparably in terms of accuracy.
Takeaway
This study shows how to combine gene expression data from different tests to get a better overall picture of gene activity.
Methodology
Factor analysis was applied to gene expression data from three microarray platforms to create unified measures.
Potential Biases
Potential bias from platforms that do not correlate well with others.
Limitations
The method may not perform well for genes with low expression levels or small dynamic ranges.
Participant Demographics
The study included 246 glioblastoma multiforme samples and 175 ovarian samples.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website