Feature gene selection method based on logistic and correlation information entropy.

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Gene expression profile data presents challenges like high dimensionality, small sample sizes, nonlinearity, and numeric types.
• Effective feature gene selection is crucial for accurate biological data analysis and classification.

Purpose of the Study:

• To develop an optimized feature gene selection method for gene expression data.
• To improve classification accuracy and reduce the dimensionality of gene expression datasets.

Main Methods:

• A hybrid approach combining logistic regression for initial screening and the Relief algorithm for candidate feature generation.
• Utilizing correlation information entropy to eliminate redundant features.
• Employing Support Vector Machine (SVM) for final classification of the selected gene subset.

Main Results:

• The proposed method successfully identifies a smaller subset of informative genes.
• Experimental results demonstrate a higher recognition rate compared to existing methods.
• The method effectively handles the complexities of high-dimensional, nonlinear gene expression data.

Conclusions:

• The integrated feature gene selection strategy is effective for gene expression data analysis.
• This method offers a robust solution for improving classification performance in genomics.
• The approach provides a valuable tool for identifying key genes in biological studies.