Identifying Important Genes for Cancer Classification
Author Information
Author(s): Su Zhenqiang, Hong Huixiao, Fang Hong, Shi Leming, Perkins Roger, Tong Weida
Primary Institution: Center for Toxicoinformatics, National Center for Toxicological Research (NCTR), U.S. Food and Drug Administration (FDA)
Hypothesis
Can a new hybrid gene selection approach improve the identification of informative genes from microarray data?
Conclusion
The VIP gene selection approach can identify additional informative genes that may provide valuable biological insights.
Supporting Evidence
- The VIP gene selection method was tested with nine publicly available microarray datasets.
- Classifiers built from VIP genes showed comparable performance to those built from p-value ranked genes.
- VIP genes may convey more biologically relevant information than p-value selected genes.
Takeaway
Scientists found a new way to pick important genes from a large list, which helps in understanding diseases better.
Methodology
A hybrid gene selection approach using bagging sampling and t-statistic for gene significance evaluation.
Potential Biases
Potential selection bias due to information leakage from training to validation phases.
Limitations
The study may be limited by the small number of samples compared to the number of genes.
Statistical Information
P-Value
0.0027
Statistical Significance
p<0.05
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website