WERFE: A Gene Selection Algorithm Based on Recursive Feature Elimination and Ensemble Strategy.

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Gene selection is crucial for micro-array data classification to identify informative genes.
• Existing algorithms often struggle with balancing performance and gene set size.
• Efficient classification requires minimizing the number of selected genes while maximizing accuracy.

Purpose of the Study:

• To propose a novel wrapper gene selection approach, WERFE, for efficient micro-array data classification.
• To develop a method that achieves high classification performance with a minimal gene set size.
• To overcome limitations of existing gene selection algorithms.

Main Methods:

• Implemented a wrapper gene selection approach named WERFE within a recursive feature elimination (RFE) framework.
• Employed an ensemble strategy integrating multiple gene selection methods.
• Assembled top selected genes from individual methods to form a final, compact gene subset.

Main Results:

• The WERFE approach demonstrated superior performance in micro-array data classification.
• Achieved state-of-the-art results by selecting a discriminative and compact gene subset.
• Effectively prioritized important genes through ensemble integration.

Conclusions:

• WERFE offers an efficient and effective solution for gene selection in micro-array classification.
• The ensemble strategy enhances the ability to identify optimal gene subsets.
• This method provides a more discriminative and compact gene selection compared to existing approaches.