Gene Selection for Multiclass Prediction
Author Information
Author(s): Xuan Jianhua, Wang Yue, Dong Yibin, Feng Yuanjian, Wang Bin, Khan Javed, Bakay Maria, Wang Zuyi, Pachman Lauren, Winokur Sara, Chen Yi-Wen, Clarke Robert, Hoffman Eric
Primary Institution: Virginia Polytechnic Institute and State University
Hypothesis
Can a two-step gene selection method improve the accuracy of multiclass disease prediction?
Conclusion
The two-step gene selection method successfully identified a small set of highly discriminative genes for improved multiclass prediction.
Supporting Evidence
- The method identified a much smaller yet efficient set of jointly discriminatory genes for diagnosing small round blue cell tumors and muscular dystrophies.
- Prediction accuracies were reported as 96.9% for small round blue cell tumors and 92.3% for muscular dystrophies.
Takeaway
Scientists found a way to pick the best genes to help tell different diseases apart, making it easier to diagnose them accurately.
Methodology
A two-step gene selection method was used, involving the identification of individually and jointly discriminatory genes, followed by evaluation using artificial neural networks and support vector machines.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website