An SVM-wrapped multiobjective evolutionary feature selection approach for identifying cancer-microRNA markers

Area of Science:

• Biochemistry
• Genetics
• Bioinformatics

Background:

• MicroRNAs (miRNAs) are crucial regulators of gene expression.
• Aberrant miRNA expression is linked to cancer development.
• Microarray data analysis is vital for identifying disease-specific biomarkers.

Purpose of the Study:

• To develop and evaluate a multiobjective feature selection method for identifying differentially expressed miRNAs in cancer.
• To pinpoint potential miRNA biomarkers for cancer detection and classification.

Main Methods:

• A multiobjective optimization approach employing Genetic Algorithms.
• Support Vector Machine (SVM) classifier used as a wrapper for feature subset evaluation.
• Application to real-life miRNA expression datasets.

Main Results:

• Successful identification of differentially expressed miRNAs associated with malignant tissues.
• Validation of identified miRNA markers through biological significance testing.
• Demonstration of the proposed method's performance on real-world data.

Conclusions:

• The proposed multiobjective feature selection method effectively identifies cancer-related miRNA markers.
• This approach aids in discovering novel diagnostic and prognostic biomarkers for cancer.
• The identified miRNA markers warrant further investigation for clinical applications.