Gene expression feature selection for prostate cancer diagnosis using a two-phase heuristic-deterministic search...

Area of Science:

• Biotechnology
• Bioinformatics
• Computational Biology

Background:

• Prostate cancer diagnosis relies on accurate identification of disease-specific biomarkers.
• Gene expression data offers a rich source for biomarker discovery but requires sophisticated analysis.
• Existing methods may struggle with high dimensionality and noise in genomic datasets.

Purpose of the Study:

• To develop and validate a novel two-phase search strategy for identifying minimal yet highly accurate gene expression biomarkers for prostate cancer.
• To enhance the precision and efficiency of biomarker discovery in complex genomic data.

Main Methods:

• A two-phase search strategy integrating statistical filtering, chaotic-tuned binary particle swarm optimization (BPSO), and cache-based sequential forward floating selection (SFFS).
• Phase 1: BPSO for optimal gene subset selection balancing gene count and classification accuracy.
• Phase 2: Cache-based SFFS for identifying the most discriminant genes from the selected subset.

Main Results:

• The proposed algorithm successfully identified a minimal set of nine informative genes from a challenging prostate cancer dataset.
• Achieved perfect classification accuracy (100%), sensitivity (100%), and specificity (100%) using the identified biomarkers.
• Demonstrated the efficacy of the two-phase strategy in pinpointing highly discriminant genes.

Conclusions:

• The developed two-phase search strategy is highly effective for identifying accurate and minimal gene expression biomarkers for prostate cancer.
• This approach offers a robust method for biomarker discovery, potentially improving early diagnosis and treatment strategies.
• The findings highlight the potential of optimized computational methods in precision oncology.