A genetic algorithm-support vector machine method with parameter optimization for selecting the tag SNPs

Area of Science:

• Genetics
• Bioinformatics
• Computational Biology

Background:

• Millions of Single Nucleotide Polymorphisms (SNPs) in the human genome offer potential for disease-gene association studies.
• High genotyping costs for millions of SNPs necessitate selecting representative subsets (tag SNPs).
• Existing tag SNP selection methods exhibit limited prediction accuracy.

Purpose of the Study:

• To develop a novel method for accurate tag SNP selection.
• To enhance prediction accuracy in identifying representative SNPs for genetic studies.

Main Methods:

• A new method, GA-SVM, integrating a genetic algorithm (GA) for tag SNP selection and support vector machine (SVM) for prediction.
• Particle Swarm Optimization (PSO) algorithm used to optimize SVM parameters (C and γ).

Main Results:

• Experimental testing on diverse datasets demonstrated superior performance of the GA-SVM method.
• The developed method achieved higher prediction accuracy in identifying tag SNPs compared to existing approaches.

Conclusions:

• The GA-SVM method with parameter optimization presents a significant advancement in tag SNP selection.
• This approach offers improved accuracy and cost-effectiveness for large-scale genetic association studies.