Feature selection algorithm based on dual correlation filters for cancer-associated somatic variants

Area of Science:

• Genomics
• Bioinformatics
• Cancer Research

Background:

• Sequencing technology has generated vast genetic information, increasing the need for human cancer analysis.
• Identifying cancer-associated variants among numerous genetic variations is crucial for understanding cancer.
• The effects of genetic variants on human cancer development are a growing area of research.

Purpose of the Study:

• To propose a novel filter-based feature selection method for extracting cancer-associated somatic variants.
• To analyze variants linked to both cancer activation and deactivation using dual correlation filters.
• To address the challenge of identifying significant variants efficiently from large datasets.

Main Methods:

• Development of a dual correlation filter-based feature selection method.
• Utilizing multiobjective optimization to simultaneously analyze activating and deactivating cancer variants.
• Employing correlation-based weighting to manage computational complexity and select significant variants.

Main Results:

• The proposed method effectively extracts cancer-associated variants.
• Dual correlation filters analyze variants related to cancer characteristics.
• Application to melanoma metastasis and breast cancer staging demonstrated classification capabilities.

Conclusions:

• The dual correlation filter-based method successfully extracts cancer-associated variants.
• The approach is effective in identifying variants linked to human cancer characteristics.
• This method contributes to advancing genetic variant analysis in cancer research.