MACOED: a multi-objective ant colony optimization algorithm for SNP epistasis detection in genome-wide association...

Area of Science:

• Genetics
• Bioinformatics
• Computational Biology

Background:

• Existing genetic-interaction detection methods in genome-wide association studies (GWAS) suffer from performance variability due to single-correlation models.
• Single-objective methods exhibit low power and high false-positive rates, particularly for complex disease models.

Purpose of the Study:

• To introduce MACOED, a multi-objective heuristic optimization methodology for robust genetic interaction detection.
• To improve upon the limitations of single-objective methods in GWAS by integrating complementary statistical approaches.

Main Methods:

• MACOED integrates logistic regression and Bayesian network methods within a multi-objective framework.
• A memory-based multi-objective ant colony optimization algorithm addresses computational complexity in high-dimensional data.
• The methodology retains non-dominated solutions across iterations for improved efficiency.

Main Results:

• MACOED demonstrated superior performance compared to existing algorithms in both simulated and real datasets.
• The method achieved higher detection power and a lower false-positive rate than single-objective optimizations.
• Experimental results confirm MACOED's computational feasibility for large-scale datasets.

Conclusions:

• MACOED offers a more powerful and accurate approach for genetic interaction detection in GWAS.
• The multi-objective strategy effectively handles complex disease models, overcoming limitations of previous methods.
• MACOED provides a computationally efficient solution for analyzing large genetic datasets.