Identifying Key Genes for Cancer Classification
Author Information
Author(s): Baker Stuart G, Kramer Barnett S
Primary Institution: National Cancer Institute, Bethesda, MD, USA
Hypothesis
Can we find classification rules that perform well with fewer genes in microarray studies?
Conclusion
Investigators should focus on classification rules that use fewer genes and select those with high frequencies of occurrence for further study.
Supporting Evidence
- The classification rules for 5 genes performed similarly to those for 20 or 50 genes.
- Desmin was identified as a key gene for classifying colon cancer.
- Zyxin and secretory granule proteoglycan genes were important for classifying leukemia.
Takeaway
This study shows that we can find important genes for cancer classification using just a few genes instead of many, making it easier to understand and use.
Methodology
The study used multiple random validation and a nearest centroid rule to analyze data from four published cancer studies.
Limitations
The study's findings may not apply to all types of microarray studies or cancer classifications.
Statistical Information
Confidence Interval
95% confidence intervals reported for AUC
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website