MACOED

MACOED applies a memory-based multi-objective ant colony optimization algorithm to detect genetic interactions in genome-wide association studies (GWAS), integrating logistic regression and Bayesian network objectives to improve detection power and control false positives between single nucleotide polymorphisms (SNPs) and disease.

Key Features:

• Memory-based multi-objective ant colony optimization: Employs a memory-based multi-objective ant colony optimization algorithm to explore combinations of SNPs for interaction detection.
• Integration of logistic regression and Bayesian network methods: Simultaneously optimizes logistic regression and Bayesian network objectives as complementary statistical paradigms.
• Retention of non-dominated solutions: Retains non-dominated solutions from past iterations to enhance navigation of large solution spaces.
• Addressing single-correlation limitations: Targets epistatic and complex interactions beyond single-correlation models between SNPs and disease.
• High-dimensional complexity management: Manages space and time complexity challenges inherent in high-dimensional genetic data analysis.
• Benchmark performance: Demonstrates higher detection power and lower false-positive rate compared with single-objective optimization and competitive algorithms on simulated and real datasets.

Scientific Applications:

• Genetic interaction detection in GWAS: Detect interactions among SNPs in genome-wide association studies to elucidate genetic architecture of traits and diseases.
• Analysis of complex disease models: Identify complex epistatic interactions across diverse disease models.
• Evaluation on simulated and real datasets: Apply to large-scale simulated and real genetic datasets for method benchmarking and interaction discovery.

Methodology:

Implements a memory-based multi-objective ant colony optimization algorithm that retains non-dominated solutions and jointly optimizes logistic regression and Bayesian network objectives to search high-dimensional SNP interaction spaces.