Improving the Performance of SVM-RFE to Select Genes in Microarray Data
2006

Improving Gene Selection in Microarray Data

Sample size: 246 publication Evidence: moderate

Author Information

Author(s): Ding Yuanyuan, Wilkins Dawn

Primary Institution: The University of Mississippi

Hypothesis

Can a modified version of Recursive Feature Elimination (RFE) improve computational efficiency while maintaining prediction accuracy in gene selection?

Conclusion

The RFE-Annealing algorithm is efficient and produces a gene set similar to that of traditional RFE.

Supporting Evidence

  • The RFE-Annealing algorithm took approximately 26 minutes for gene selection compared to 58 hours for RFE.
  • Both RFE and RFE-Annealing achieved around 98%-100% accuracy on test data.
  • RFE-Annealing was found to be more stable than the original RFE in most cases.

Takeaway

This study created a faster way to pick important genes from a large list, making it easier to find out if someone is sick.

Methodology

The study compared the performance of RFE and RFE-Annealing algorithms on three gene expression datasets using Support Vector Machines.

Limitations

The study's results may not generalize to all types of gene expression data or other machine learning contexts.

Participant Demographics

The study involved patients with various types of cancer, including lung carcinomas and pediatric Acute Lymphoblastic Leukemia.

Digital Object Identifier (DOI)

10.1186/1471-2105-7-S2-S12

Want to read the original?

Access the complete publication on the publisher's website

View Original Publication