A novel hybrid algorithm based on Harris Hawks for tumor feature gene selection.

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• High-dimensional gene expression data present challenges in identifying cancer-specific genes.
• Accurate selection of informative feature genes is crucial for effective cancer classification.

Purpose of the Study:

• To propose a novel, efficient, and accurate three-stage hybrid gene selection method for cancer classification.
• To enhance the identification of key genes contributing to tumor development and progression.

Main Methods:

• A three-stage hybrid approach combining a variance filter, extremely randomized tree, and Harris Hawks optimization (VEH).
• Stage 1: Variance filter to pre-select genes based on a threshold.
• Stage 2: Extremely randomized tree for eliminating irrelevant genes.
• Stage 3: Harris Hawks algorithm for optimal feature gene subset selection.

Main Results:

• The VEH method achieved 100% classification accuracy for gastric cancer, acute lymphoblastic leukemia, and ovarian cancer.
• An average classification accuracy of 95.33% was observed across diverse cancer types.
• VEH demonstrated superior performance compared to other advanced feature selection algorithms based on multiple evaluation criteria.

Conclusions:

• The proposed VEH method is highly effective for feature gene selection in cancer classification.
• VEH offers significant advantages in accuracy and efficiency over existing methods for analyzing high-dimensional gene expression data.
• This approach holds promise for improving diagnostic and prognostic tools in oncology.