It has recently been shown that cancer genes (oncogenes) tend to have heterogeneous expressions across disease samples. So it is reasonable to assume that in a microarray data only a subset of disease samples will be activated (often referred to as outliers), which presents some new challenges for statistical analysis. In this paper, we study the multi-class cancer outlier differential gene expression detection. Statistical methods will be proposed to take into account the expression heterogeneity. Through simulation studies and application to public microarray data, we will show that the proposed methods could provide more comprehensive analysis results and improve upon the traditional differential gene expression detection methods, which often ignore the expression heterogeneity and may loss power. Supplementary information can be found at http://www.biostat.umn.edu/∼baolin/research/orf.html.
|Number of pages
|Computational Biology and Chemistry
|Published - Apr 2007
Bibliographical noteFunding Information:
This research was partially supported by a research grant from the Minnesota Medical Foundation.
- Cancer gene activation heterogeneity
- Differential gene expression detection
- False discovery rate
- Robust regression